Quality assurance of PASADENA hyperpolarization for ¹³C biomolecules

Abstract

Object

Define MR quality assurance procedures for maximal PASADENA hyperpolarization of a biological ¹³C molecular imaging reagent.

Materials and methods

An automated PASADENA polarizer and a parahydrogen generator were installed. ¹³C enriched hydroxyethyl acrylate, 1-¹³C, 2,3,3-d₃ (HEA), was converted to hyperpolarized hydroxyethyl propionate, 1-¹³C, 2,3,3-d₃ (HEP) and fumaric acid, 1-¹³C, 2,3-d₂ (FUM) to hyperpolarized succinic acid, 1-¹³C, 2,3-d₂ (SUC), by reaction with parahydrogen and norbornadiene rhodium catalyst. Incremental optimization of successive steps in PASADENA was implemented. MR spectra and in vivo images of hyperpolarized ¹³C imaging agents were acquired at 1.5 and 4.7 T.

Results

Application of quality assurance (QA) criteria resulted in incremental optimization of the individual steps in PASADENA implementation. Optimal hyperpolarization of HEP of P = 20% was achieved by calibration of the NMR unit of the polarizer (B₀ field strength ± 0.002 mT). Mean hyperpolarization of SUC, P = [15.3 ± 1.9]% (N = 16) in D₂O, and P = [12.8 ± 3.1]% (N = 12) in H₂O, was achieved every 5–8 min (range 13–20%). An in vivo ¹³C succinate image of a rat was produced.

Conclusion

PASADENA spin hyperpolarization of SUC to 15.3% in average was demonstrated (37,400 fold signal enhancement at 4.7 T). The biological fate of ¹³C succinate, a normally occurring cellular intermediate, might be monitored with enhanced sensitivity.

Introduction

PASADENA (Parahydrogen and synthesis allow dramatically enhanced nuclear alignment [1,2] is unique in its ability to achieve an extremely high degree of hyperpolarization at liquid state temperatures within seconds and at relative low cost. This is achieved by catalytic addition of molecular para-hydrogen to a precursor molecule, while spin order is manipulated consecutively for the generation of net polarization on a designated nucleus, in this case ¹³C. Nuclear spin polarization depicts the alignment of spins exposed to a magnetic field, and corresponds to the fraction of all spins, that contribute to NMR signal (Eq. 1):

$$P=\frac{N(\uparrow )-N(\downarrow )}{N(\uparrow )+N(\downarrow )}.$$ (1)

At ambient temperature, the thermal energy exceeds the transition energy of the spin states by orders of magnitudes, leaving only ~1–100 ppm of all spins contributing to the overall NMR signal. This is described mathematically by the Bolzmann distribution (Eq. 2):

$$P_{\text{Boltzmann}}=tanh\left(\frac{\gamma \hslash B_0}{2k_{B}T}\right)\approx \frac{\gamma \hslash B_0}{2k_{B}T},$$ (2)

where γ is gyromagnetic ratio characteristic for a nucleus of interest, ħ is the Plank constant divided by 2π, B₀ is the field strength, k_B is the Boltzmann constant and T is the temperature.

Hyperpolarization techniques hold the potential to increase the polarization to the order of unity, which is an enhancement of 4–5 orders of magnitude (η_observed) with respect to the thermal polarization originating from the magnetic field (Eq. 3):

$${\eta }_{\text{observed}}=\frac{P_{HP}}{P_{\text{Boltzmann}}}.$$ (3)

The polarization achieved in total, P_HP, provides a measure for the efficiency of the hyperpolarization technique, independent of the specific system for detection. It is predominantly used in this paper (Eq. 4):

$$P_{\text{HP}}={\eta }_{\text{observed}}·P_{\text{Boltzmann}}.$$ (4)

Heteronuclear PASADENA hyperpolarization employs the spin order added to the precursor molecule by addition of parahydrogen to form spin polarization on a third, designated nucleus, such as, but not limited to, ¹³C or ¹⁵N [1–3].

The system of two spin 1/2 particles of parahydrogen was predicted 1924 by Born and Heisenberg [4], and experimentally measured three years later by Bonhoeffer and Harteck [5]. The singlet state holds spin zero (Eq. 5), while the triplet states has spin one (Eq. 6).

$$\begin{array}{l}\mid 0,0\rangle =1/\sqrt{2}(\mid \uparrow \downarrow \rangle -\mid \uparrow \downarrow \rangle )\\ \mid 1,1\rangle =\mid \uparrow \uparrow \rangle \end{array}$$ (5)

$$\begin{array}{l}\mid 1,-1\rangle =\mid \downarrow \downarrow \rangle \\ \mid 1,0\rangle =1/\sqrt{2}(\mid \uparrow \downarrow \rangle +\mid \uparrow \downarrow \rangle )\end{array}$$ (6)

The r.f. pulse sequence to form scalar order in the ¹³C nucleus in a parahydrogenated molecule is described in Goldman et al. [6]. Since the evolution of the spin system is affected by the J-couplings characteristic to the molecular environment, the spin order transfer sequence is specific to each individual molecule.

The method has previously been restricted, as early examples were non-biological molecules, soluble only in acetone. A water-soluble molecule which was toxic at the doses employed, and confined to the vasculature after injection in vivo was employed in a demonstration of ¹³C angiography [7,8]. A further selection of hyperpolarizable molecules is described by Kuhn and Bargon [9]. A metabolizable molecule, ¹³C sodium acetylenedicarboxylate which is converted to ¹³C succinate by two hydrogenation steps, has also been reported from this laboratory [10]. In a recent report, we described hyperpolarization of 1-¹³C in succinate [11].

Succinate is an intermediate in the tricarboxylic acid cycle and other significant metabolic pathways, including gluconeogenesis and the recently defined HIF1α oncogene [12,13]. With the advent of imaging reagents with the potential to broaden the use of parahydrogen hyperpolarization in biology, comes the need to standardize. In a previous manuscript, we described the equipment necessary to perform individual steps of the experiment: PASADENA generation, chemistry and hardware of the hydrogenation reaction and the MR electronics for the spin order transfer [14]. Here we present the user with procedures of quality assurance necessary for the setup and optimization of polarization yield, and introduce standard operational protocols developed for day-to-day hyperpolarization experiments in vivo.

Theoretically, PASADENA is realized in three steps (Fig. 1a–c). In practice, a number of factors determine the overall success of hyperpolarization. First, the quality of parahydrogen, the source of the spin order, limits the level of hyperpolarization. Second, the addition of parahydrogen to the ¹³C-enriched precursor molecule, the catalyst and the reaction conditions determine whether or not the target molecule is formed efficiently and in such a way that parahydrogen spin order is available to the third step, the manipulations of the spins and transfer of polarization to ¹³C (or ¹⁵N). The efficacy of these manipulations is determined by the accuracy with which the J-couplings of the target molecule are known, and how well the designated manipulations are applied to the spins by r.f. hardware. We describe a technique whereby highly reproducible operation of the PASADENA polarizer can be achieved and using succinic acid, 1-¹³C, 2,3-d₂ (SUC), demonstrate its efficacy in vivo.

Fig. 1.

Chemistry and spin physics of the PASADENA experiment. (a) (left) molecular parahydrogen ( ) is added to the precursor ( ) by catalytic hydrogenation; (center) manipulations of the spins of the molecule by spin order transfer r.f. pulse sequence, involving the ²J_CHa,³ J_CHb,³ J_HaHb couplings; (right) net spin polarization is formed on ¹³C. (b) Hydrogenation of precursor FUM forms the product SUC. (c) Hydrogenation reaction cycle of Rhodium based catalyst: the norbornadiene moeity is hydrogenated and becomes labile and combines with the bisphosphine ligand to form the catalyst. The PASADENA precursor attaches itself to this active catalyst and hydrogenation occurs by addition [15,16]

Materials and methods

Table 1 establishes a protocol for PASADENA hyperpolarization. For convenience, the materials prerequisite (column A) we describe in Methods while installation, calibration (column B) and the experimental routine (column C), all of which lead to Quality Assurance guidelines, are presented in “Results”.

Table 1.

Protocol for PASADENA hyperpolarization

A. Prerequisites	B. Installation and calibration	C. Quality assurance routine
Parahydrogen (>95%) Precursor molecule(s) Hydrogenation catalyst PASADENA polarizer	Polarizer fluid control system Calibration of the low field NMR system Determine center frequency Calibration of the low field NMR system: adjust flip angles 1H, 13C J couplings for target molecule B0 optimization B1 optimization	Flush the polarizer Flush delivery system Prepare MRI scanner Check temperature Check gas pressure (N2, pH2) Check B0 value Check r.f. amplitudes of B1 pulses

(A) materials required for successful hyperpolarization, (B) tasks required to establish PASADENA hyperpolarization, optimize the operating equipment and (C) quality assurance routines in daily use are summarized under these headings For A, see “Methods”; for B and C, see “Results”

Prerequisites for PASADENA

Parahydrogen

Commercially available ultra-pure hydrogen (Gilmore, South El Monte, USA) was catalytically converted to parahydrogen (pH₂) by slow passage over granular hydrous ferric oxide (IONEX-type O–P catalyst; Molecular Products Inc., Lafayette, CO, USA). After the gas was converted to para-hydrogen it was stored in 7 L aluminum cylinders at room temperature at a pressure of 33 bar. The quality of pH₂ was determined to be >97% by high resolution NMR [7]. Each batch was used within 7 days, with no measurable decrease in the yield.

Imaging reagent precursors

For initial calibration, a previously validated PASADENA reagent, hydroxyethyl acrylate, 1-¹³C, 2,3,3-d₃ (HEA), was prepared as an aqueous solution. 2–5 mg HEA (Isotec Sigma–Aldrich, USA) was dissolved in a small volume of phosphate buffer, pH 7.0, mixed with the catalyst solution and de-aerated using a vacuum line (Schlenk). A second PASA-DENA reagent [11] introduced for further calibration and biomedical studies; fumaric acid, 1-¹³C, 2,3-d₂ (FUM); (Cambridge Isotope Laboratories, Andover, MA, USA) was dissolved in a small volume of phosphate buffer at pH 2.9. Low pH was necessary for well-defined J-couplings in the resulting hydrogenation product, SUC (as described further below).

Hydrogenation catalyst

A unique catalyst developed for the purpose of hydrogenation which preserves the desired spin order [15,16] was freshly prepared by mixing two components. The bisphosphine ligand, 1,4-bis-[(phenyl-3-propane sulfonate) phosphine] butane disodium salt (Q36333, Isotec, OH, USA), was dissolved in H₂O/D₂O to yield 2.5–3.0 mmol/L concentration followed by removal of oxygen using vacuum and nitrogen connected via a manifold. The rhodium catalytic moiety was then introduced to the reaction mixture under nitrogen (N₂) atmosphere as a solution of bis(norbornadiene)rhodium (I) tetrafluoroborate (catalog number 45-0230, CAS 36620-11-8, Strem Chemicals, MA, USA) in acetone with 5% molar excess of bisphosphine ligand with respect to rhodium. The resulting solution was vigorously shaken and acetone was removed under vacuum at room temperature. Excess of bio-phosphine is necessary for complete removal of rhodium.

The imaging reagent solution was added to the catalyst and alternately exposed to inert gas (N₂) and vacuum. The resulting mixture contained 1–3 mmol/L FUM and 2.0–2.5 mmol/L catalyst concentrations in 50 mmol/L pH 2.9 phosphate buffer. The aqueous mixture of catalyst and molecular precursor was prepared fresh, prior to each hyperpolarization procedure. The completed PASADENA solution of precursor and catalyst was drawn into a 20 mL plastic syringe and connected to valve 2 of the PASADENA polarizer, for injection of the desired amount of imaging reagent precursor for each experiment (3.5 mL unless otherwise noted). When prepared in these proportions, hydrogenation of the precursor was carried to completion and no residual precursor (HEA or FUM) was detected by ¹³C NMR.

PASADENA polarizer

The very detailed description of the central instrument required are provided in an accompanying manuscript [14].

Other methods

Determination of J-couplings of the PASADENA agents

As indicated in Fig. 1 and described in some detail by Bowers and Weitekamp [1,2], Natterer and Bargon [17] and more recently by Goldman et al. [6,18] significant transfer of spin order from parahydrogen to the third nucleus is achieved through design and implementation of a spin transfer sequence which is then applied to the mixture of ¹³C reagent and rhodium catalyst. In order to achieve ¹³C hyperpolarization by r.f. radiation, the spin order transfer (SOT) sequence should be tailored to the J-couplings of the newly added hydrogen and the nucleus to-be-polarized (²J_CHa,³ J_CHb,³J_HaHb) in the product molecule. For the experiments described, two SOT sequences designed for HEP [6,18] and SUC [11], respectively, were programmed on the synthesizer incorporated into the polarizer and delivered as r.f. pulses to the B₁ coil of the low field NMR.

Quantification of hyperpolarization

A number of strategies have been employed to quantify the extent of hyperpolarization [19]. In the present paper, the following convention was applied. The signal enhancement (η_observed) achieved by hyperpolarization was quantified in respect to the signal of a thermally polarized sample (100% ethanol, 188 mM natural abundance ¹³C at each site (Eq. 7)). The signal intensities were determined by numerical integration (xWINNMR software, Bruker Biospin, Germany):

$${\eta }_{\text{observed}}=\frac{P_{\text{HP}}}{P_{\text{Boltzmann}}}=\frac{I_{HP}}{I_{\text{ref}}}·\frac{c_{\text{ref}}}{c_{\text{HP}}}.$$ (7)

where I_HP, I_ref, c_HP and c_ref are integral intensities and molar concentrations of hyperpolarized and reference samples, respectively.

T₁ decay of hyperpolarization

The decay of ¹³C polarization caused by longitudinal relaxation (T₁) of the hyperpolarized sample during the delivery to the detection system was determined (see “Results”). This allows for the determination of the nascent signal enhancement (Eq. 8), and nascent level of achieved polarization (Eq. 9):

$${\eta }_{\text{observed}}={\eta }^{t=0}·exp\left(-\frac{\text{delivery}\text{time}}{T_1}\right)$$ (8)

$$P_{\text{HP}}^{t=0}={\eta }^{t=0}·P_{\text{Boltzmann}}.$$ (9)

Unless otherwise indicated, $P_{\text{HP}}^{t=0}$ and η^t=0 are reported in the work presented, using a ¹³C polarization of P_Boltzmann = 0.00041% or 4.1 ppm at 298 K and 4.7 T. T₁ values for the PASADENA reagent concerned in calibration and quality assurance (QA) were determined directly. A series of small angle (α ≈ 8°) pulse-and-collect experiments was employed to probe the decay of the magnetization of the hyperpolarized agents (small excitation angle approximation, SEA). The T₁ values were extracted by the fit of (Eq. 10) to the data points, taking into account the loss of magnetization caused by the excitation pulses:

$$I(t)=I_0·{\text{e}}^{-\left(\frac{t}{T_1}\right)}·{\text{e}}^{-\left(\frac{1-cos(\alpha )t}{TR}\right)}.$$ (10)

Results

Installation and calibration of the PASADENA experiment

Seven steps are indicated (Table 1, Column B) for the set-up and calibration of the polarizer. Together these constitute a Quality Assurance routine which ensures maximal hyperpolarization and optimal reproducibility for each PASADENA reagent.

Step 1: polarizer fluid control system

An automated fluid control system delivers reagents to the reaction chamber. The steps of the hyperpolarization experiment, including the fluid control system were controlled by custom software (LabView platform, National Instruments, Austin, TX, USA) able to load a r.f. pulse file, and to store the sequence of timings and events separate file. The following procedure was optimized to produce hyperpolarized agent in less than 1 min: (1) flush the system with N₂ gas (14 bar), (2) load the aqueous solution of precursor and catalyst in the N₂ path, (3) fill the previously flushed reactor with 10 bar pH₂ and pressurize N₂ path, (4) close pH₂ path, inject precursor solution in pH₂ atmosphere while applying ¹H r.f. saturation pulses (to maintain the singlet state of hydrogen attached to molecules), (5) apply the SOT, (6) deliver the hyperpolarized solution to the detector (eg. small animal NMR or clinical MRI scanner) employing residual pressure. The reader can refer to [14] for a description of the valves function.

Step 2: calibration of the NMR system of the polarizer

An initial calibration of the NMR system of the polarizer is necessary to perform accurate r.f. manipulations on the spins of the PASADENA agent in the reactor. Since the r.f. electronics and coil were designed for transmission only, a second, external high-field NMR system was employed for the detection (7 T Mercury spectrometer, Varian, Palo Alto, USA). A sample of saturated 1-¹³C sodium acetate CH_3–¹³COONa in D₂O was pre-polarized at high field for 3 min, delivered to the polarizer for a ¹³C inversion pulse of arbitrary length, followed by delivery back to the high field unit, where the resulting magnetization was detected by a 90° pulse-acquisition sequence.

Step 3: determine center frequency of NMR system of polarizer

To center the frequency, a ¹³C sample was subjected repeatedly to a “prepolarize_Varian –pulse_polarizer – delay – pulse-break acquisition_Varian” experiment while the strength of the low-field in the polarizer was incremented by 0.1 mT, and the r.f. was kept constant: the optimal field for the 18 kHz carbon pulse was found to be ~1.76 mT at the position of the Hall sensor of the gauss meter (Fig. 2). Once the optimum field for the carbon frequency was found, the corresponding proton frequency was calculated (75 kHz).

Fig. 2.

Experiment to center the frequency of the low field MR unit of the polarizer. A constant ¹³C excitation pulse was applied to a sample of saturated 1-¹³C sodium acetate in D₂O, followed by the transfer (~14 s) and a pulse-acquisition detection in a high field spectrometer. This procedure was repeated for eight settings of the static magnetic field to determine the optimum (1.763 mT). The scale refers to the offset from this optimum

Step 4: flip angle calibration

For the calibration of the flip angles of the r.f. excitation pulses (B₁), the previous protocol for the combination of low-field pulse and high field detection (Fig. 3) was employed. At constant B₀, the width of the r.f. pulse given in the polarizer was varied for ¹³C and ¹H frequencies independently. The acquired data was fitted using a T₁ corrected cosine function, providing an estimate for the B₁ fields.

Fig. 3.

¹³C and ¹H flip angle calibration of the low field MR unit in the polarizer. A ¹³C or ¹H square excitation pulse was applied to a sample of saturated 1-¹³C sodium acetate in D₂O, followed by the transfer (~14 s) and a pulse-acquisition detection in a high field spectrometer. This procedure was repeated with 22 and 31 different pulse widths for ¹³C, ¹H, respectively, at constant amplitudes (¹H 31.8 V, ¹³C 9.6 V). Inversion pulses were determined to be 157 μs (¹H) and 550 μs (¹³C), respectively

Step 5: determination of J-couplings

While the J-coupling constants for HEP employed in calibration of the polarizer have previously been reported [10], the relevant J-couplings for succinate were unknown and were determined by computations employing GAMMA simulations [20] and confirmed by experiment [11]. The ¹³C spectra (Fig. 4) were acquired using a saturated (100 mM) aqueous solution of succinic acid (natural abundance, Isotec, Sigma Aldrich, USA) at a 14 T high resolution spectrometer (Varian, H/C/N probe, 256 acquisitions) without decoupling.

Fig. 4.

Determination of the relevant J-couplings for the hyperpolarization of succinate. (a) ¹³C NMR spectrum of natural abundance succinic acid at pH 7.4 (14 T Varian spectrometer, 256 acquisitions, H/C/N probe, no decoupling). (b) ¹³C NMR spectra of carboxyl (1, 4-¹³C, ~185 ppm) and methylene (2, 3-¹³C, ~35 ppm) groups at pH 7.4. Note the line broadening of the former. (c) Simulations (red) and measured ¹³C NMR spectra of carboxyl and methylene groups at pH 2.95. Note that at pH 2.95 improved resolution of the line splittings in carboxyl resonances are observed. The J-couplings extracted of the simulated spectrum were ²J_Cha = −7.15 Hz, ³J_CHb = 5.82 Hz and ³J_HaHb = 7.41 Hz. These values were employed to calculate the spin order transfer sequence for succinate

At pH 7.4 (Fig. 4b), substantial line broadening attributed to chemical proton exchange was observed on the carboxyl group (~185 ppm) which prohibited the resolution of line splitting by J-couplings. All experiments were therefore performed at pH 2.95 ≪ pKa (Fig. 4c).

For the simulations, the spectra of a ¹³C spin coupled to two protons (C, Ha, Hb) were calculated under the effect of Zeeman interaction and indirect couplings (²J_CHa,³ J_CHb,³J_HaHb).

The simulations were performed iteratively varying the J-couplings until the best correspondence of the simulated to the experimental spectra (Levenberg–Marquart least-square fit) was found (²J_CHa = −7.15 Hz, ³J_CHb = 5.82 Hz and ³J_HaHb = 7.41 Hz).

These constants were employed to tailor the spin order transfer sequence to the target molecule, SUC.

Step 6: optimization of spin order transfer: B₀ calibration with hyperpolarization

Introducing parahydrogen and substituting a PASADENA reagent and catalyst, acceptable levels of hyperpolarized signal were achieved as detected in the high-field spectrometer. The optimization of B₀ to center the frequency was repeated over a smaller range, recording the yield of hyperpolarization at each field (Fig. 5). Maximal polarization of $P_{\text{HP}}^{t=0}=21\%$ (nominal ¹³C signal enhancement η^t=0 ≊ 50,000 at 4.7 T) was found at 1.763 mT.

Fig. 5.

Calibration of the center frequency of the low field MR unit. Maximal polarization achieved in this experimental set-up was 21%, somewhat lower than previously reported [18]. The PASADENA experiment was conducted at different strengths of the static field using HEP (0.1 mM). This was very sensitive to variations of the static field: Half of the polarization was lost at a field offset of ±0.009 mT (full width at half maximum (FWHM) = 0.018 mT). The amplitudes and durations of the B₁ pulses of the SOT used are shown in Fig. 3

As expected, hyperpolarization was extremely sensitive to deviations from the precise low magnetic field applied during spin order transfer. At ±0.009 mT from the optimal field, the polarization declined by 50% (FWHM = 0.018 mT). Marked variations persisted in the polarization yield at the same field strength, which were attributed to remaining variables in r.f. amplifier performance, reagent preparation technique and, because signal decays rapidly before detection, to prolonged and variable sample delivery times to the spectrometer. Delivering hyperpolarized sample and triggered detection in the NMR scanner were further optimized. The polarizer was moved to the in vivo suite, where it was positioned carefully close to a small-animal horizontal bore, high field MR (7.63 m ±0.5 cm distance, 4.7 T Bruker MRI scanner). An automated system was implemented, whereby the hyperpolarized solution was driven by remaining gas pressure in the reactor, through polyethelene (PE) tubing from the polarizer to an NMR-tube in a dual-tuned ¹H/¹³C solenoid coil in the MR scanner, where after a precise interval (33 s) acquisition was triggered. The solution was then drained to permit successive experiments without coil repositioning. Together, these measures proved to be essential for the reliability of the level of achieved polarization. The modifications implemented improved the stability of the B₀ and provided stronger B₁, allowing for better r.f. performance and less sensitivity of hyperpolarization of HEP to off-resonant B₀, as demonstrated in Fig. 6 (left).

Fig. 6.

Impact of r.f. amplifier power to hyperpolarization. (left) New r.f. hardware allowed for increase in r.f. power (previously 25V/16V in Fig. 5, now 25V/50V (for carbon, hydrogen, respectively): 3.1-fold increase) and less sensitivity of hyperpolarization to off-resonance frequency (previously FWHM = 0.018 mT, now FWHM = 0.062 mT, 3.4-fold increase). Adaptation of the spin order transfer sequence allowed for hyperpolarization of SUC, which was conducted ten times at different strengths of the static field (right)

Step 7: B₁ optimization with hyperpolarization

With the goal of in vivo application, the next group of calibrations were performed with the non-toxic, water soluble PASADENA biomolecule SUC (Fig. 6, right).

The hyperpolarization yield was incrementally increased by further optimization of the individual pulse widths of ¹³C, ¹H in the SOT sequence, at constant B₀ (Fig. 7). First, the pulse width of the ¹H pulses in the SOT was varied (Fig. 7a). The optimum was determined to be at 115 μs (50 V). Next, keeping the ¹H pulse width constant, the experiment was repeated while varying the ¹³C pulse widths in the SOT sequence (Fig. 7b). Each experiment was conducted eight times. The optimum ¹³C pulse width was 230 μs (25 V). Fig. 7c demonstrates the multidimensional character of this optimization.

Fig. 7.

Optimization of SUC hyperpolarization yield by two-dimensional optimization of the B₁ flip angles. First, the PASADENA experiment was conducted 8 times. Each time the pulse width of the ¹H pulses in the SOT was varied (a). The optimum was determined to be at 115 μs (50 V). Next, the experiment was repeated under variation of the ¹³C pulse widths in the SOT (b). The optimum found was 230 μs (25 V). (c) demonstration of the multidimensional character of this optimization

The maximal polarization achieved with 1-¹³C succinate under these conditions was 18%.

T₁ of short-lived hyperpolarization

Using these QA methods, we defined two important properties of PASADENA for a biologically appropriate molecular imaging reagent. First, T₁ of hyperpolarized MR signal is critical in defining duration of enhanced ¹³C (or ¹⁵N) MR signal after delivery. We determined the maximum polarization achieved in PASADENA as relevant for quantitative comparison with competing reagents and hyperpolarization techniques. Secondly, the effect of D₂O prolonging the life-time of polarization is a general property, and of value for most hyperpolarization techniques. The T₁ decay time for HEP was 70 s in D₂O, as compared to 50 s in H₂O (pH 7, B₀ = 4.7 T).

The relaxation constant of hyperpolarized SUC was determined to be T₁ = 27 ± 3] s (N = 3, pH 3, B₀ = 4.7 T) in H₂O solvent (at a concentration of 1–3 mM of SUC, 2.5 mM of catalyst). T₁ = [40 ± 2] s (N = 4, pH 3, B₀ = 4.7 T) was significantly prolonged when measured in D₂O (Fig. 8). Preliminary results indicate a further increase in measured T₁ to ~50 s, when the hyperpolarized reagent–catalyst mixture was buffered to physiological pH 7 at 4.7 T (not shown).

Fig. 8.

T₁ measurement of hyperpolarized SUC in D₂O (T₁= [40 ± 2] s) and in H₂O (T₁= [27 ± 3] s). The decaying polarization was probed by small angle pulse-acquisition experiments in a high field system (4.7 T)

Quality assurance routine

A check-list is provided (Table 1, column C) which is recommended to be performed as quality assurance each day. These seven steps were found to ensure a high level of cleanliness of the entire polarizer circuit and reaction chamber as well as accurate function of the electronics necessary for delivery, circulation and SOT-spin physics.

Quantification and reproducibility of polarization

Repeated hyperpolarization experiments of SUC in D₂O were carried out to determine the level and stability of the polarization produced by the PASADENA polarizer. A representative ¹³C spectrum of hyperpolarized SUC is demonstrated in Fig. 9a. Peak amplitude was quantified by comparison with the spectrum of ethanol (Fig. 9b), using the equations described in “Methods”. Polarization of ¹³C succinate in D₂O achieved 19%, in this representative experiment, corresponds to an overall enhancement of η^t=0 ≈ 49,000-fold at 4.7 T.

Fig. 9.

Quantification of polarization: (a) ¹³C spectra of SUC (1.95 mM) hyperpolarized to 19%, corresponding to an enhancement of ~47,000 at 4.7 T and (b) unlabeled thermally polarized (P = 4.1 × 10⁻⁴%) ethanol (¹³C concentration 188 mM per site natural abundance)

To define the reproducibility of polarization, 28 further studies were performed. Note that since each experiment takes only 2–3 min, with allowance for rinsing and reloading of the polarizer, the automated polarizer permitted as many as ten PASADENA hyperpolarizations each hour.

Greater reproducibility and higher absolute polarization were achieved after B₁ calibration and when the procedure was performed in D₂O (Table 2). Hyperpolarization. $P_{\text{HP}}^{t=0}=[15.3\pm 1.9]\%$ was reproducibly achieved in 16 hyperpolarization experiments on SUC, in D₂O. This corresponds to a relative enhancement of η^t=0 = 37,400± 4,600 at 4.7 T. The intra day variability was found to be $P_{\text{HP}}^{t=0}=[16.5\pm 3.3]\%$, N=3, $P_{\text{HP}}^{t=0}=[15.9\pm 2.0]\%$, N = 4, $P_{\text{HP}}^{t=0}=[14.8\pm 1.3]\%$, N = 5, $P_{\text{HP}}^{t=0}=[14.5\pm 0.9]\%$. Detailed results are presented graphically in Fig. 10. More relevant to the conduct of experiments was the degree of hyperpolarization at the point of measurement. As expected with a mean delay of 33 s for delivery and T₁ = 27 s to 40 s, enhancement was significantly lower, $P_{\text{HP}}^{t=33s}=[6.4\pm 0.8]\%$, N = 16, in D₂O (enhancement η^t=0 ≈ 37,400 × [6.5/15.3] = 15,900-fold at the point of measurement).

Table 2.

Reproducibility and performance of the polarizer, D₂O

(A) Number	(B) $P_{\text{HP}}^{t=0}$ (%)	(C) $P_{\text{HP}}^{t=33s}$ (%)	(D) c (SUC) (mM)	(E) c (Rh) (mM)
1	15.5	6.5	1.71	2.2
2	20.1	8.4	1.71	2.2
3	13.8	5.8	1.71	2.2
Mean (N = 3)	16.5 ± 3.3	6.9 ± 1.3
4	17.1	7.2	1.4	2.2
5	15.9	6.7	1.4	2.2
6	17.5	7.3	1.4	2.2
7	13.0	5.5	0.96	2.2
Mean (N = 4)	15.9 ± 2.0	7.1 ± 0.3
8	14.9	6.2	2.93	2.1
9	15.4	6.4	2.93	2.1
10	16.3	6.8	1.24	2.2
11	14.7	6.2	1.24	2.2
12	12.7	5.3	1.24	2.2
Mean (N = 5)	14.8 ± 1.3	6.5 ± 0.3
13	15.3	6.4	1.59	2.2
14	14.3	6.0	1.95	2.2
15	15.1	6.3	1.59	2.2
16	13.3	5.6	2.07	2.2
Mean (N = 4)	14.5 ± 0.9	6.2 ± 0.2
Total (N = 16)	15.3 ± 1.9	6.4 ± 0.8

Four different sets of experiments of hyperpolarization of SUC in D₂O were performed as indicated (Fig. 10). From left to right: (A) number of experiment; (B) level of nascent polarization; (C) level of polarization after 33 s; (D) concentration of SUC in mM; (E) concentration of Rh–catalyst complex in mM. Level of polarization after 33 s (C) is equivalent to that available at the point of delivery to an experimental animal

Fig. 10.

Reproducibility of hyperpolarization. $P_{\text{HP}}^{t=0}=[15.3\pm 1.9]\%$ was achieved in 16 hyperpolarization experiments on SUC, in D₂O conducted on successive days. Details are provided in Table 2

No effect of the concentration of SUC on the level of hyperpolarization was found over a narrow range of concentrations (0.96–2.93 mM in D₂O; Table 2) or (1.65–2.89 mM in water; Table 3). Although it is anticipated that injection of small amounts of D₂O will be harmless, hyperpolarization was also determined in aqueous solution (Table 3). The average polarization achieved in H₂O, calculated from T₁ to be that at the point of production, was $P_{\text{HP}}^{t=0}=[12.8\pm 3.1]\%$ on SUC (2–5 mM) for 12 experiments in three series, not significantly different (P > 0.05) from results in D₂O ( $P_{\text{HP}}^{t=33s}=[5.4\pm 1.3]\%$ in water).

Table 3.

Reproducibility and performance of the polarizer, H₂O

(A) Number	(B) $P_{\text{HP}}^{t=0}$ (%)	(C) $P_{\text{HP}}^{t=33s}$ (%)	(D) c (SUC) (mM)	(E) c (Rh) (mM)
1	16.2	6.8	1.85	2.2
2	19.0	8.0	1.85	2.2
3	17.5	7.3	1.85	2.2
Mean (N = 3)	17.6 ± 1.4	7.4 ± 0.6
4	10.1	4.2	2.89	2.2
5	13.1	5.5	2.89	2.2
6	12.4	5.2	2.89	2.2
7	12.2	5.1	2.89	2.2
Mean (N = 4)	12.0 ± 1.3	5.0 ± 0.6
8	9.8	4.1	1.65	2.2
9	10.2	4.3	1.65	2.2
10	11.6	4.9	1.65	2.2
11	11.8	5.0	1.65	2.2
12	10.6	4.4	1.65	2.2
Mean (N = 5)	10.8 ± 0.9	4.5 ± 0.4
Total (N = 12)	12.9 ± 3.1	5.4 ± 1.3

Three sets of experiments of hyperpolarization of SUC in H₂O were performed as indicated. For details, see Table 2

Efficacy of hyperpolarized 1-¹³C succinate in vivo

Figure 11 demonstrates, in a representative ¹³C image, overlaid on a standard proton image, the efficacy of hyperpolarized ¹³C succinate for in vivo imaging. After rapid ante-grade injection into the common carotid artery, hyperpolarized ¹³C succinate appeared in the anatomical distribution coincident with the rat brain, as outlined by ¹H MRI.

Fig. 11.

In vivo ¹³C sub-second image (0.3 s) of a rat brain, acquired 9 s after close-arterial injection of 1 ml, 25 mM hyperpolarized SUC (in color). The ¹³C image was overlaid on a coronal ¹H fast gradient echo image with matching field-of-view (FOV) and slice location acquired prior to infusion to provide anatomical correlation (3D FIESTA, TR/TE = 6.3/3.1 ms, 5 × 5 × 5 mm³ spatial resolution, FOV = 220 mm/320 mm, 44 phase encoding steps/64 readout points, respectively)

Discussion

A reliable method for the hyperpolarization of a biomolecule for routine applications in bio-medical research is presented in this work. The performance is demonstrated on SUC, a ¹³C PASADENA metabolic agent with the potential to diagnose brain cancer [10]. This molecule was polarized to an average of $P_{\text{HP}}^{t=0}=[15.3\pm 1.9]\%$ in 16 experiments, corresponding to a 37,400-fold enhancement at 4.7 T, with a maximum enhancement of η^t=0 ≈ 49,000-fold.

The robust design of the polarizer, reflected in the achieved levels and reproducibility of polarization, qualifies (after reagent-sterility is confirmed) for routine bio-medical/clinical application. However, the theoretical maximum of polarization is not reached, and it is worthwhile to consider the remaining limitations of the experiment.

Three major factors govern both the reproducibility and level of polarization achieved by PASADENA: (A) The primary source of spin order, parahydrogen, (B) the hydrogenation of the precursor molecule, and (C) the manipulation of the spins in the hydrogenated molecule to generate a net polarization (by r.f. Spin order transfer sequence, SOT).

1. The parahydrogen is routinely enriched to >97%, thereby guaranteeing an excellent source of spin order. It proved to be important to allow a minimum of 6 h for the cryo-system to cool down, prior to the generation of parahydrogen.

2. It was noted that an inert atmosphere was a key aspect in the preparation of the PASADENA precursor solution. This is attributed to the sensitivity of the Rhodium (Rh) based catalyst to oxygen and moisture. Therefore, bis(norbornadiene)rhodium (I) tetrafluoroborate should be used fresh and with minimum air and light exposure during the chemical preparation of catalyst solution. An automated setup under inert atmosphere would be desirable to minimize the exposure of the Rhodium catalyst to atmospheric oxygen, accelerate the preparation process and reduce experimental variability (currently under development in our laboratory). The duration of the hydrogenation reaction may hold further potential for improvements, as preliminary results indicate. This may be attributed to the progress of the hydrogenation reaction, and the loss of spin order state of molecular and bound parahydrogen while the reaction is going on. However, the hydrogenation reaction was deemed to be complete to >99% within 4 s under 10 bar pH₂ and 62°C, which is supported by the fact that no precursor signal was found in high field NMR spectrum of the hydrogenated sample.

3. This leaves the spin manipulations by r.f. to explain why the theoretical value of unity, >90%, ¹³C polarization was not achieved. The pattern of the spin manipulations tailored to the molecule, and how it is realized by NMR electronics, are essential to achieve hyperpolarization. While the first requires knowledge of the coupling constants of the target molecule, the latter is mostly governed by experimental imperfections of the setup.

The static field of the 4.7 T MR unit used for the detection of hyperpolarization was found to perturb the relatively low field in PASADENA polarizer. At a distance of (7.6 ± 0.1) m, the stray field of the 4.7 T adds a ~0.1 mT horizontal component to the vertical 1.7 mT field of the polarizer. This effect is manifested in a complete loss of hyperpolarization if the polarizer is moved in to close proximity of the non-shielded magnet. While this challenge may not persist in the proximity of modern, actively shielded high field magnets, active (field compensation gradients) or passive shielding for the PASADENA polarizer are potential solutions under investigation in our laboratory. To avoid thermal drift of the B₀ field, a minimum of 4 h was allowed for the polarizer to reach operational temperature (62°C).

The sensitivity of the hyperpolarization towards fluctuations in B₀ could be reduced by the application of even stronger B₁ pulses. However, possible limitations arise: first, large (with respect to B₀) B₁ fields could make the pulse sequence perform sub-optimally as conventional NMR theory was used for the development (approximation B₀ ≫ B₁). Second, at strong B₁, interaction of ¹³C and 1H has to be taken into account: There is increased cross-talk between the two, and the pulses applied at one Larmor frequency may affect the spins at the other. Moreover, ²H nuclei present in the SUC may be affected by the r.f. pulses, too, a matter not yet investigated in detail. Preliminary simulations suggest that the cross-talk between the two channels can be as large as 9% under the conditions of the present study. These challenges may be significantly alleviated by (1) conducting the hyperpolarization experiment at a higher field (~10 mT) and/or (2) design of shaped pulses for selective excitation at low field.

Conclusion

A PASADENA polarizer is presented, capable of achieving reproducible polarization of a test reagent, custom-designed for application in bio-medical research. The protocols employed for the production of hyperpolarized substrate are provided, along with a detailed description of the optimization and quality assurance procedures. The performance of the apparatus is demonstrated on succinate, an intermediate of the Krebs-TCA cycle and many other vital processes. In 16 experiments, performed over 4 different days, an average polarization of $P_{\text{HP}}^{t=0}=[15.3\pm 1.9]\%$ was achieved. The fully automated set up can be operated by a single person, and allows continuous production of hyperpolarized agent within seconds at liquid state temperatures. Limitations to the hardware are discussed, and potential improvements outlined.

While the unique chemical specificity of nuclear magnetic resonance holds great potential in bio-medical research, it is limited by the inherently low polarization (order of 10⁻⁵) to in vivo detection of major moieties (~mM). Routinely available PASADENA hyperpolarization will encourage exploration of metabolic events involving metabolites in the micro or even nanomolar range.