Real-time metabolic imaging with dissolution DNP — watching the Warburg effect happen in vivo
Hyperpolarized ¹³C MRI begins not in the scanner but in a chemistry lab: a small quantity of [1-¹³C]-labeled pyruvic acid, doped with a stable free radical, is flash-frozen into a glassy solid. This unassuming frozen pellet is the raw material that, after polarization, will let radiologists watch cellular metabolism happen in real time inside a living patient.
The choice of substrate and isotope-labeling position is deliberate and central to the whole technique. Pyruvate is labeled specifically at carbon-1 (the carboxyl carbon) rather than at C2 or C3, because the C1 position has a comparatively long longitudinal relaxation time T1 in vivo (~43-65 seconds at 3T) — long enough to survive the transport from polarizer to patient to scanner and still retain usable signal.
Pyruvate itself is an ideal metabolic reporter because it sits at a central metabolic branch point: it can be reduced to lactate (via LDH, using NADH), transaminated to alanine (via ALT), or oxidized to acetyl-CoA and CO2/bicarbonate (via the pyruvate dehydrogenase complex, PDH). Each of these fates is detectable as a distinct, chemically shifted ¹³C peak in the same spectroscopic acquisition — giving a simultaneous readout of glycolytic flux, amino acid metabolism, and oxidative metabolism from one injection.
Critically, pyruvate is a small, endogenous, non-toxic molecule already present in human blood at low millimolar concentrations, and the injected dose (typically raising blood pyruvate by only 3-8 fold transiently) does not meaningfully perturb normal physiology — this is a tracer technique, not a pharmacological intervention.
Dynamic nuclear polarization requires a source of highly polarizable, unpaired electron spins whose enormous polarization (electrons have a magnetic moment ~658 times larger than a proton's, and ~2,600 times larger than ¹³C's) can be transferred to the target nucleus. This is provided by doping the frozen pyruvate glass with a persistent organic free radical — most commonly the trityl radical OX063 (tris[8-carboxyl-2,2,6,6-tetra(hydroxyethyl)benzo[1,2-d:4,5-d']bis(1,3)dithiole-4-yl]methyl sodium salt), typically at ~15 mM concentration.
Trityl radicals are chosen because their narrow EPR linewidth allows efficient, selective microwave excitation, and — critically for clinical translation — the radical can be removed from the dissolved pyruvate solution via filtration before injection (using a silica-based radical filter integrated into GMP polarizer systems), keeping free radical exposure to the patient below safety thresholds.
The pyruvate/trityl mixture is flash-frozen (often by direct injection into liquid helium or rapid immersion) to form a disordered, glassy solid rather than a crystalline one — glassy disorder is important because it distributes the radical and substrate molecules relatively homogeneously and avoids phase separation that would reduce polarization transfer efficiency.
Dynamic nuclear polarization (DNP), first adapted for dissolution and in vivo imaging by Ardenkjær-Larsen and colleagues in 2003, is the enabling technology of the entire field. By irradiating the frozen doped sample with microwaves in a strong magnetic field at cryogenic temperature, the huge polarization of unpaired electrons is transferred to neighboring ¹³C nuclei — boosting nuclear polarization by four to five orders of magnitude relative to thermal equilibrium.
At thermal equilibrium, nuclear spin polarization follows the Boltzmann distribution: P = tanh(ħγB/2kT), which for ¹³C (low gyromagnetic ratio γ) at clinical field strengths and body temperature yields a minuscule polarization of roughly 0.0001% — meaning that of every million ¹³C nuclei, only about one contributes a net detectable signal above its antiparallel-spin twin.
Electrons, with a magnetic moment ~658-fold larger than protons, have a correspondingly much higher thermal polarization even at modest field and temperature. DNP exploits microwave-driven electron-nuclear cross-relaxation processes — primarily the solid effect and cross effect (the dominant mechanism for trityl radicals at high field) — to transfer this large electron polarization to nearby ¹³C nuclei via the hyperfine coupling and dipolar interactions between unpaired electron spins and nuclear spins.
Microwave irradiation is tuned close to the electron Larmor frequency (~94 GHz at 3.35T), driving simultaneous electron-nuclear spin flips that progressively "pump" polarization from the electron bath into the nuclear spin system. Because nuclear T1 at 1K is extremely long (hours), polarization can accumulate to far above thermal levels over the 1-3 hour build-up period, reaching typical values of 20-40% (i.e., a near-complete alignment of nuclear spins, compared to ~1-in-a-million alignment at body temperature).
The resulting enhancement — commonly quoted as 10,000 to 50,000-fold over thermal polarization — is what makes ¹³C MRI feasible at all: natural-abundance or even isotopically-enriched ¹³C signal is normally far too weak for real-time in vivo imaging, but hyperpolarization briefly makes ¹³C nearly as MR-visible as ¹H.
Translating DNP from a research technique into a reproducible clinical tool required purpose-built, GMP (Good Manufacturing Practice)-compliant hardware. The GE Healthcare SPINlab polarizer, used at UCSF, is the first FDA-cleared clinical DNP polarizer and enables sterile, single-use, cassette-based sample preparation compliant with pharmaceutical manufacturing standards.
Key engineering features include:
• Fully automated, closed fluid pathway — the pyruvate/trityl mixture is polarized, dissolved, and quality-tested without operator contact, minimizing contamination risk • Multi-sample capacity — allows sequential polarization of multiple doses per day • In-line quality control — measures final polarization level, pH, temperature, residual trityl radical concentration, and pyruvate concentration before the dose is released for injection, all within regulatory limits (e.g., residual trityl <3 μmol/L) • Sterility and pyrogen testing integrated into the automated workflow to meet the same safety bar as any injectable pharmaceutical
This infrastructure investment — turning a physics technique requiring liquid helium and microwave engineering into a same-day clinical workflow — has been essential to enabling multi-site clinical trials of hyperpolarized ¹³C imaging.
The single most delicate step of the entire pipeline is dissolution: the frozen, hyperpolarized pellet must be rapidly converted into a warm, physiological, injectable liquid — without destroying the fragile nonequilibrium nuclear polarization that took over an hour to build. This is accomplished in under two seconds using a jet of superheated, pressurized solvent.
Dissolution DNP works by rapidly melting and dispersing the frozen hyperpolarized pellet using a jet of pressurized, superheated water or buffered solvent (typically containing NaOH, tris buffer, and EDTA to neutralize the strongly acidic pyruvic acid and chelate any trace paramagnetic metal ions that would otherwise catalyze rapid polarization loss).
The solvent is heated to roughly 180°C and kept pressurized (to prevent boiling) until it is released through a narrow line directly onto the frozen sample inside the polarizer's dissolution chamber. The thermal energy of the superheated jet melts and dissolves the sample within 1-2 seconds, producing a physiological-temperature (~37°C after further dilution), neutral-pH, hyperpolarized pyruvate solution.
This process is inherently lossy: nuclear T1 during the brief high-temperature, high-thermal-motion dissolution transient is short, and typically 30-40% of the polarization achieved in the solid state is lost during dissolution and subsequent transfer, filtration (radical removal), and quality control — meaning the injected polarization is typically in the 20-35% range even when the solid-state polarizer achieved 40%+.
From the moment of dissolution, hyperpolarization begins decaying irreversibly back toward thermal equilibrium with a characteristic T1 relaxation time. Because this decay cannot be paused or reversed, every subsequent step — quality control testing, transfer to the injection syringe, patient injection, vascular transit, cellular uptake, and image acquisition — must occur within a race against the clock defined by T1.
For [1-¹³C]pyruvate, in vivo T1 is approximately 43-65 seconds at clinical field strengths (values are somewhat longer at lower field and shorter at higher field, and also tissue-dependent). This means:
• Injection must begin within about 15-25 seconds of dissolution to preserve clinically useful signal • The total imaging window is effectively only about 60-90 seconds — after roughly 2-3 T1 periods (~90-130s), signal has decayed to a small fraction of its initial value and is no longer usable • Unlike conventional contrast agents that can be imaged for many minutes, hyperpolarized agents are fundamentally single-shot, time-critical probes — there is no way to "wait and re-image" once polarization is spent; a fresh dose (and a fresh 1+ hour polarization cycle) would be required to repeat the measurement
This short, non-renewable imaging window is the defining engineering constraint of the entire field, driving the use of rapid, undersampled acquisition schemes (spiral, EPI, spectral-spatial excitation with small flip angles to conserve remaining magnetization across repeated readouts).
Because each RF excitation pulse irreversibly consumes some of the finite, non-renewable hyperpolarized magnetization (unlike thermally polarized spins, which recover via T1 relaxation between pulses), hyperpolarized acquisitions use small flip angles (often 5-20°) and variable flip angle schedules to "spend" the available magnetization economically across the full dynamic time series.
Once injected (typically as an 8 mL dose delivered over ~10-12 seconds via a peripheral IV line, dosed at approximately 0.43 mL/kg body weight of a ~230-250 mM pyruvate solution), the hyperpolarized bolus must travel through the vasculature, cross the capillary endothelium, and enter the cytosol of target cells before any metabolic conversion can be imaged.
Pyruvate, like lactate, is a monocarboxylate anion that cannot freely diffuse across the lipid bilayer at physiological pH and must be actively shuttled across the cell membrane by monocarboxylate transporters (MCTs), a family of proton-linked symporters. MCT1 is broadly expressed across most tissues and mediates baseline pyruvate/lactate exchange; MCT4 is more specifically upregulated in highly glycolytic tissues, including many cancers, as part of the broader metabolic reprogramming that supports aerobic glycolysis.
Because MCT expression itself varies between tumor types and can be upregulated by hypoxia-inducible factor (HIF-1α) signaling in hypoxic tumor regions, pyruvate delivery and uptake (reflected in the "pyruvate signal" or "kTRANS"-like influx metric in hyperpolarized MRI analysis, distinct from the DCE-MRI Ktrans) is itself informative — tumors with poor pyruvate delivery may reflect poor perfusion or reduced transporter expression, complicating interpretation of a low lactate signal (which could mean either low LDH flux or simply low substrate delivery).
Unlike gadolinium-based DCE-MRI where the imaging window is minutes long, hyperpolarized pyruvate delivery must be captured within the first 20-40 seconds after injection, overlapping substantially with the T1-decay-limited acquisition window. This creates a practical tension: too early an acquisition catches vascular pyruvate before conversion has occurred; too late an acquisition misses the peak conversion dynamics as polarization itself has decayed.
Most clinical protocols therefore use dynamic, time-resolved acquisitions (one low-flip-angle spectroscopic or imaging readout every 2-5 seconds for ~60-90 seconds after injection) that capture the full temporal evolution of both pyruvate delivery and lactate (and other metabolite) build-up, allowing kinetic modeling analogous in spirit — though mechanistically distinct — to the pharmacokinetic curve-fitting used in DCE-MRI.
The central clinical application of hyperpolarized ¹³C pyruvate imaging is direct, real-time visualization of the Warburg effect — the preference of cancer cells for aerobic glycolysis (converting glucose-derived pyruvate to lactate even in the presence of adequate oxygen) rather than oxidative phosphorylation. The rate of ¹³C label exchange from pyruvate to lactate, quantified as the apparent rate constant kPL, directly reports this altered metabolic phenotype.
Lactate dehydrogenase (LDH) catalyzes the reversible interconversion of pyruvate and lactate: Pyruvate + NADH + H⁺ ⇌ Lactate + NAD⁺. Because the reaction is fully reversible and near-equilibrium in most cells, and because the total cellular pyruvate + lactate pool is vastly larger than the injected hyperpolarized tracer bolus, the observed hyperpolarized ¹³C label exchange behaves as a pseudo-first-order chemical exchange process, described by simple two-site exchange kinetics:
d[Lactate*]/dt = kPL·[Pyruvate*] − kLP·[Lactate*] − [Lactate*]/T1,lac
where the asterisk denotes hyperpolarized-labeled pool, kPL is the apparent forward pyruvate-to-lactate exchange rate constant (s⁻¹), and the equation is coupled to a parallel decay equation for pyruvate* including its own T1 and outflow terms.
The key insight is that kPL is not simply the LDH enzymatic rate constant in isolation — it depends on the product of LDH-A expression/activity, the size and redox state (NADH/NAD⁺ ratio) of the intracellular lactate pool, and the total lactate pool size (since exchange with a larger unlabeled pool dilutes the observable signal per unit time). Elevated LDH-A expression, elevated NADH availability from high glycolytic flux, and expanded intracellular lactate pools — all hallmarks of the Warburg phenotype — combine to produce a higher observed kPL in aggressive tumors.
The Warburg effect, first described by Otto Warburg in the 1920s, refers to cancer cells' preference for glycolysis followed by lactate fermentation even under normoxic (oxygen-sufficient) conditions — a metabolic signature so consistent across cancer types that it underlies the entire field of FDG-PET oncologic imaging. Hyperpolarized ¹³C pyruvate MRI provides a complementary, radiation-free window onto the same phenomenon, with the added ability to distinguish glycolysis (lactate), oxidative metabolism (bicarbonate via PDH), and transamination (alanine) simultaneously.
Multiple converging factors elevate the pyruvate-to-lactate conversion rate specifically in aggressive tumor tissue:
• HIF-1α-driven transcriptional upregulation of LDH-A, even under normoxia in many tumors (pseudohypoxia driven by oncogenes like MYC and loss of tumor suppressors like VHL or PTEN) • Increased expression of MCT1/MCT4 transporters facilitating both pyruvate uptake and lactate export, sustaining high glycolytic throughput • Elevated cytosolic NADH availability from high glycolytic flux, providing the reducing equivalents LDH requires • Expanded total intracellular lactate pool size in glycolytic tumors, which — counterintuitively — can further amplify the observed hyperpolarized exchange signal in some kinetic regimes
This biological cascade produces a graded relationship between tumor aggressiveness and observed kPL or lactate-to-pyruvate ratio that has been validated across preclinical models and, increasingly, in human clinical trials — positioning hyperpolarized pyruvate MRI as a potential grading and treatment-response biomarker distinct from and complementary to anatomic imaging.
Hyperpolarized [1-¹³C]pyruvate MRI progressed from first preclinical demonstrations (Golman et al., 2006) to first-in-human imaging in prostate cancer within just seven years — a remarkably rapid translational timeline for a fundamentally new imaging modality — and has since expanded into brain, breast, renal, and cardiac applications.
The first-in-human study of hyperpolarized [1-¹³C]pyruvate MRI (Nelson et al., Science Translational Medicine, 2013) imaged 31 patients with biopsy-confirmed prostate cancer at UCSF prior to prostatectomy, using the SPINlab polarizer system. The study demonstrated that the hyperpolarized lactate-to-pyruvate ratio was significantly elevated in regions of biopsy-confirmed higher-grade cancer compared to normal peripheral zone tissue, and that the spatial extent of elevated lactate correlated with tumor extent determined on whole-mount pathology after surgery.
Critically, the study established the safety and feasibility of the entire dissolution DNP pipeline in humans — no dose-limiting toxicities were observed, and the injected pyruvate dose was well below levels expected to cause any pharmacological effect, confirming that hyperpolarized MRI functions as a true tracer technique rather than a pharmacological intervention. This trial catalyzed rapid international expansion of the technology to multiple academic centers.
Follow-up multi-site studies extended hyperpolarized pyruvate imaging to active surveillance monitoring in low-risk prostate cancer patients, where rising lactate-to-pyruvate ratios on serial imaging could flag early disease progression before conventional MRI or PSA changes — a potential non-invasive alternative to repeated biopsies.
Building on the prostate experience, clinical trials of hyperpolarized [1-¹³C]pyruvate MRI have expanded to several other tumor types:
• Brain tumors (glioma): distinguishing tumor progression from treatment-related changes (pseudoprogression, radiation necrosis) — a major unmet clinical need where conventional contrast-enhanced MRI is often ambiguous; elevated lactate labeling has shown promise for identifying viable, metabolically active tumor • Breast cancer: assessing early treatment response to neoadjuvant chemotherapy, with reductions in lactate-to-pyruvate ratio observed within days of treatment initiation in responding tumors, analogous to the early Ktrans changes seen with anti-angiogenic therapy in DCE-MRI • Renal cell carcinoma: differentiating clear cell RCC (typically VHL-mutant, highly glycolytic) from other renal tumor subtypes and normal renal parenchyma based on distinct lactate conversion kinetics • Metastatic and hepatic applications remain areas of active investigation, complicated by the liver's naturally high baseline metabolic activity
Beyond oncology, hyperpolarized [1-¹³C]pyruvate MRI offers a unique window into cardiac substrate metabolism, distinguishing the heart's reliance on glycolysis/lactate production versus oxidative phosphorylation via the pyruvate dehydrogenase (PDH) pathway — a distinction with direct relevance to heart failure, diabetic cardiomyopathy, and ischemia.
In the heart, pyruvate can follow three fates simultaneously detectable in a single hyperpolarized acquisition:
• Lactate (via LDH) — anaerobic/glycolytic flux, elevated during ischemia or in a "fetal gene program" metabolic shift seen in heart failure • Bicarbonate/CO2 (via PDH, the rate-limiting step of oxidative glucose metabolism) — the flagship marker of oxidative metabolic capacity; PDH flux is markedly suppressed in diabetic and insulin-resistant hearts due to PDH kinase (PDK4) upregulation • Alanine (via ALT, alanine transaminase) — reflects glycolytic flux and amino acid nitrogen exchange, less central to cardiac studies than lactate/bicarbonate
Human cardiac hyperpolarized pyruvate studies (first reported ~2016 onward) have demonstrated feasibility of measuring myocardial PDH flux non-invasively, with early trials in diabetic patients and post-myocardial-infarction patients showing reduced bicarbonate signal (reduced oxidative flux) consistent with known pathophysiology — opening a path toward a novel, radiation-free cardiac metabolic stress test.
| Product | Indication | Trial Design | Key Result |
|---|---|---|---|
| [1-¹³C]Pyruvate | Prostate, brain, breast, renal cancer; cardiac | LDH → lactate; PDH → bicarbonate; ALT → alanine | Most validated probe; FDA-cleared polarizer (SPINlab) |
| [1-¹³C]Fumarate | Necrosis / cell death imaging | Fumarase converts fumarate → malate only in necrotic cells (membrane rupture) | Direct marker of acute cell death, treatment response |
| [U-¹³C, ¹⁵N₂]Glutamine | Glutaminolysis-dependent tumors | Reports glutamine-driven anaplerosis in MYC-driven cancers | Complements Warburg (glycolysis) readout with a second oncogenic pathway |
| [1-¹³C]α-ketoglutarate / others | IDH-mutant glioma metabolism | Probes altered TCA cycle flux and 2-HG oncometabolite production | Direct link to a defined oncogenic mutation (IDH1/2) |