Quasi-continuous production of highly hyperpolarized carbon-13 contrast agents every 15 seconds within an MRI system

Magnetic resonance imaging relies on weak thermal polarization, limiting sensitivity and restricting clinical imaging largely to abundant nuclei and molecules. Hyperpolarized contrast agents (HyCAs) overcome these sensitivity limits, but current production methods such as dissolution dynamic nuclear polarization (dDNP) are complex, costly, and slow (typically ≥1 hour per dose), hindering widespread application. Parahydrogen-induced polarization (PHIP) offers a faster, less expensive alternative, with hydrogenative PHIP enabling rapid creation of hyperpolarization and SABRE enabling continuous polarization without consumption of the precursor. The study aims to demonstrate quasi-continuous, rapid production of 13C-hyperpolarized agents within an MRI system using SAMBADENA (synthesis amid the magnet bore), achieving duty cycles much faster than T1 relaxation to enable repeated injections or continuous infusion and prolonged observation windows for hyperpolarized MRI.

The paper reviews prior approaches for generating HyCAs: dDNP systems (e.g., Spinlab, spin-aligner) improved throughput but still require long preparation and recovery times. Hydrogenative PHIP has achieved duty cycles of minutes with >10% 13C polarization using pre-prepared batches or PHIP polarizers, enabling in vivo metabolic imaging of pyruvate and fumarate. Continuous hyperpolarization has been explored via excess precursor and continuous pH2 supply, microfluidic lab-on-chip platforms, bubble-free dissolution, and heterogeneous catalysis, with SABRE particularly suited for continuous operation and recent demonstrations of continuous X-nucleus polarization up to a few percent. SAMBADENA previously enabled in-bore production and immediate use of agents, reducing polarization losses during transfer. Collectively, these works motivate developing an integrated, rapidly cycling PHIP system capable of producing repeated doses faster than relaxation times.

Experimental platform: Hyperpolarization was implemented directly within a 7 T preclinical MRI (Bruker Biospec 70/20) equipped with a dual-resonant 1H–13C volume transmit/receive coil (10 cm length, 7.2 cm ID). The MRI pulse program provided transistor–transistor logic (TTL) outputs for precise timing (12.5 ns resolution), controlling a Teensy 3.5 microcontroller that actuated electromagnetic valves via relays to handle gases (N2, pH2) and liquid flows. A custom polysulfone reactor (inner volume ~4 mL; PSU 1000/2000 variants) with an integrated water-heated jacket (90–95 °C circulating water) enabled stable high temperature and pressure operation. The inner chamber had four ports for solution and gases; hot water flowed through an outer jacket at 2 mm spacing via insulated tubing and a pump. Though experiments used ≤20 bar, the reactor was designed and verified for much higher pressures (simulated up to 273 bar yield limit; practically tested up to 50 bar). The reactor was mounted on a motorized slider for accurate positioning in the magnet isocenter.

Fluidics: Five electronically switched magnetic valves and five one-way check valves directed flow. A fluid trap protected valves from contamination. Catalyst–precursor solution resided in a 10 mL syringe connected by ETFE tubing (1/16" OD, 0.5 mm ID, ~1 m length). Spent solution was removed via a modified syringe and guided out of the magnet through non-magnetic tubing to a collection container. Syringes S1 (fill) and S2 (eject) were manually actuated while all valve and gas operations were automated by the MRI-controlled microcontroller.

Hyperpolarization cycle (Rapid-PHIP/SAMBADENA): The cycle time was 15 s per batch: approximately 5 s to fill the reactor with pre-heated aqueous catalyst–precursor solution, 5 s for hydrogenation and spin order transfer (SOT), and 5 s to eject the polarized product. Reaction conditions were maintained at 80 °C and ≥15 bar pH2. Hydrogenation used parahydrogen addition to the precursor followed by SOT to 13C via pulse sequences: phINEPT+ for hydroxyethyl-[1-13C]propionate-d3 (HEP) and Goldman’s sequence for [1-13C]succinate-d2 (SUC). A dual-tuned 1H/13C coil executed SOT and detected the hyperpolarized 13C signal. The MRI pulse program orchestrated valve timing, gas pressurization, and RF sequences for precise synchronization.

Agents and batches: Two agents with established SAMBADENA compatibility were used: HEP and SUC. Each 15 s, a 700 µL batch containing 5 mM of target agent was produced (10 consecutive cycles for HEP; 9 for SUC). Reaction parameters: catalyst 2 mM; hydrogenation 5 s at 15 bar pH2 and 80 °C.

Quantification: 13C polarization was quantified against a thermally polarized reference (carbonyl resonance of 5 mL neat acetone, natural abundance 13C = 1.1%) at 7 T. Assumed 100% hydrogenation for polarization calculation (later analysis suggests ~90% actual hydrogenation under the chosen conditions). Signal enhancements relative to thermal equilibrium were reported.

Optimization studies: Parameter sweeps assessed dependence of 13C polarization on temperature (T), parahydrogen pressure (PpH2), catalyst concentration (1–4 mM), precursor concentration (CHEA, 5–80 mM for hydroxyethyl [1-13C] acrylate-d3, HEA), and hydrogenation time (th). Payload (precursor concentration × polarization) versus CHEA was fitted by a mono-exponential saturation, indicating pH2-limited kinetics at fixed catalyst. Hydrogenation-time dependence was modeled by a two-process kinetics expression incorporating hydrogenation (Thydr) and relaxation (Trelax), yielding fitted parameters and reacted fractions at given th.

Temperature adaptation for in vivo use: Post-reactor cooling was evaluated by guiding 5 cm of the ejection catheter through a 16 °C water reservoir. Ten consecutive 500 µL doses at 100 µL s−1 were cooled to 32 ± 2 °C at the catheter tip, demonstrating rapid temperature adjustment compatible with in vivo administration.

• Quasi-continuous production: Hyperpolarized 13C agents were produced every 15 s within the MRI bore using SAMBADENA with automated, precisely synchronized fluidics and RF control.
• Polarization levels: Mean 13C polarization for HEP was 19 ± 1% (N=10 cycles) and for SUC was 1.7 ± 0.2% (N=9 cycles), corresponding to ~31,700-fold and ~2,800-fold signal enhancements at 7 T, respectively.
• Underestimation factors: Assuming complete hydrogenation likely underestimated polarization; ~90% hydrogenation was achieved, implying a correction factor of ~1.11. Using 100% pH2 (vs. 85% used) would add another factor of ~1.25.
• Variability: Rapid-PHIP showed larger polarization variability (±20% relative) compared to single-batch SAMBADENA (±10%), attributed to varying effective volumes in the reactor due to fast injections and spillover during hydrogenation.
• Kinetic optimization: Optimal hydrogenation time th ≈ 5 s maximized polarization under tested conditions (80 °C, 15 bar pH2, 2 mM catalyst, CHEA 5 mM). A two-parameter kinetics model fitted the polarization vs. th with parameters: Thydr = 1.6 ± 0.9 s, Trelax = 16 ± 6 s, Pmax = 27 ± 5%, t0 = 1.0 ± 0.5 s. Estimated reacted fractions: 92 ± 12% at 5 s and 99 ± 1% at 10 s.
• Rate limitation: Payload saturation with increasing HEA concentration and minimal dependence on catalyst (1–4 mM) implicated pH2 availability/solubility as the rate-limiting factor.
• Environmental control: Elevated temperature and higher pH2 pressure improved yields. Cooling strategy reduced ejected solution temperature to 32 ± 2 °C at the catheter tip for 500 µL doses at 100 µL s−1.
• Throughput: Four boli per minute (10 boli in 2.5 min) were demonstrated, far exceeding dDNP duty cycles (hours) and prior PHIP batch cycles (minutes).

The study demonstrates that integrating hydrogenative PHIP within the MRI bore (SAMBADENA) and tightly synchronizing chemistry with RF control enables production of hyperpolarized 13C agents on timescales (15 s) much shorter than typical T1 relaxation, thereby mitigating SNR losses and enabling new acquisition paradigms such as repeated bolus injections, continuous infusion, and signal averaging over extended periods. Compared with dDNP, the approach is orders of magnitude faster and drastically less expensive, potentially broadening access. The observed performance is constrained principally by hydrogen solubility and reaction kinetics rather than parahydrogen supply, indicating clear engineering levers (pressure, solvent choice, temperature) for further gains. Ensuring purity and safety for in vivo use remains essential; however, rapid purification and catalyst removal strategies reported in the literature could be integrated into subsequent cycles during ongoing polarization. The method appears adaptable to a broader suite of PHIP-SAH-derived metabolic agents (e.g., fumarate, pyruvate, acetate, lactate), though solvent compatibility and catalyst handling will require modifications. Overall, producing agents faster than their decay opens practical routes to higher SNR and dynamic metabolic monitoring in preclinical settings.

This work introduces a rapid, quasi-continuous PHIP methodology integrated within an MRI system that produces 13C-hyperpolarized agents every 15 s with high polarization (≈19% for HEP and ≈2% for SUC). A custom high-pressure, high-temperature reactor with MRI-driven fluidic control achieved fast hydrogenation and efficient spin-order transfer, enabling throughput far exceeding dDNP and prior PHIP implementations. The approach paves the way for serial injections, continuous infusion, and signal averaging to enhance hyperpolarized MRI, while reducing cost and complexity. Future research should focus on increasing pH2 availability (higher pressure, improved gas dissolution), exploring solvents with higher H2 solubility (e.g., chloroform, acetone) especially for PHIP-SAH agents, optimizing SOT sequences for higher transfer efficiency, integrating rapid purification/catalyst removal, and validating metabolic agents suitable for safe in vivo translation.

• Purity and safety: The produced solutions contain catalyst and unreacted precursor; in vivo use requires rapid purification or catalyst capture, which was not implemented here.
• Polarization variability: Greater cycle-to-cycle variation in polarization (±20% relative) was observed, partly due to variable effective volumes and fast fluid handling.
• Kinetic/solubility limits: Hydrogen solubility in water and reaction kinetics limit payload at higher precursor concentrations; performance likely improves with higher pH2 pressures or solvents with higher H2 solubility.
• Scope: Demonstrations were limited to two agents (HEP, SUC) in aqueous solution; extension to PHIP-SAH metabolites will need solvent compatibility and hardware adjustments.
• Quantification assumptions: Polarization estimates assumed complete hydrogenation and used 85% pH2 enrichment, likely underestimating true polarization; correction factors were discussed but not implemented in reported means.
• Manual steps: Syringe operations (fill/eject) were manual, which may introduce variability; full automation could improve reproducibility.