The brute force approach is perhaps the simplest to understand but currently the least effective. It involves placing the sample to be polarized in a very high magnetic field at extremely low temperatures. In these conditions the sample will naturally approach a high level of polarization because the polarization achieved by a sample increases with the magnitude of the static magnetic field and decreases with the temperature. However the brute force approach is limited because the size of the polarization levels that can be achieved are relatively small and the sample can take a long time to approach the high polarization level. There are also practical difficulties in returning the sample to the liquid state for use in an in vivo experiment [5].

Optical pumping of the noble gases, particularly helium (³He) and xenon (¹²⁹Xe), has generated a great deal of interest in recent years due to its direct applicability to the imaging of the lungs [4]. The sample to be polarized is mixed with either an alkali-metal vapor or a vapor of metastable atoms. The mixture is then irradiated with circularly polarized light from a laser source, which is absorbed by the alkali or metastable atoms, increasing the polarization of their electrons [5]. This high level of polarization is then passed on to the nuclei of the noble gas through collisions. MR images of the lungs generated following the inhalation of these hyperpolarized noble gases can then be used to detect changes in ventilation, perfusion, and lung microstructure associated with diseases such as chronic obstructive pulmonary disease (COPD) and asthma [4].

PHIP, also referred to as PASADENA, utilizes a highly polarized form of molecular hydrogen, known as parahydrogen, to increase the polarization of a specific nucleus, typically carbon (¹³C) or nitrogen (¹⁵N), within a chosen molecule [2]. The first step in the PHIP technique is to generate high levels of parahydrogen. Molecular hydrogen is normally found in two different forms, orthohydrogen and parahydrogen, in a 3:1 ratio. At low temperatures, and in the presence of a suitable catalyst, the relative abundance of parahydrogen can be increased to nearly 100%. The parahydrogen produced is then involved in a hydrogenation reaction, in the presence of a second catalyst, that produces the molecule to be hyperpolarized. Finally, the high polarization of the parahydrogen is transferred to the ¹³C or ¹⁵N nuclei by either cycling the magnetic field or applying a sequence of RF pulses [5]. The PHIP technique can rapidly generate extremely high levels of polarization (> 40%) using relatively simple and inexpensive apparatus [2]. However, the technique is limited by the specificity of the chemical reaction required and the need to remove the hydrogenation reaction catalyst before injection.

The final technique, and the focus of the rest of this article, is dynamic nuclear polarization (DNP) [3, 6]. As with PHIP, the DNP technique is used to enhance the polarization of a specific nucleus (again typically ¹³C or ¹⁵N) within a particular molecule. It requires the mixing of the molecule to be hyperpolarized with a source of free electrons, known as a radical. The mixed sample is then placed in a high magnetic field (typically 3.35 T) and rapidly frozen in liquid helium, reducing the sample temperature to approximately 1 K. In these conditions, the free electrons are nearly 100% polarized (ie, all aligned in the same direction as the static magnetic field), whereas the nuclear spins of the sample are still relatively poorly polarized. The high electron polarization can then be transferred to the nuclear spins through the irradiation of the sample with microwave energy at a specific frequency, which is related to the magnetic field strength and the atomic properties of the nuclei and radical within the sample. Further details of the processes involved in this polarization transfer are more thoroughly described in the review by Comment et al. [3].

Despite the high level of polarization that can be achieved, the biological application of DNP has been limited by the need for the hyperpolarization process to take place in the solid state. This limitation was removed by the development of the dissolution DNP process [6], whereby the highly polarized solid sample is rapidly brought into the liquid state, retaining a large proportion of the enhanced polarization. The process, first described by Ardenkjaer-Larsen and colleagues in 2003, requires the rapid melting of the hyperpolarized sample with a bolus of superheated liquid and generates an injectable sample that can be used as an in vivo MR contrast agent [6].

With hyperpolarized ¹³C-labeled tracer molecules, the large SNR and the fact that there is virtually no background signal generates an extremely high contrast-to-noise ratio (CNR), which is well suited to angiographic applications [14]. As such, the use of hyperpolarized agents for angiography has been widely explored [15, 16], with particular focus being given to imaging of the coronary [17] and pulmonary [18] arteries. The long transverse relaxation times of ¹³C molecules favor the use of single shot imaging techniques (e.g. trueFISP, EPI) [1] and so high temporal resolutions (< 300 ms) have been demonstrated. The lack of background signal has also led to the use of projection imaging techniques with reasonable in-plane resolutions (~ 1 mm). However, the imaging of ¹³C-labeled molecules places large demands on the imaging gradients of the MRI system because the gyromagnetic ratio of ¹³C is ¼ that of ¹H. This leads to the need for more powerful gradient amplifiers and, when combined with rapid imaging techniques, may place limits on the ultimate spatial resolution that is achievable for angiographic applications.

CMR assessments of myocardial perfusion are usually based on the first pass of a gadolinium-based contrast agent through the heart. The contrast agent shortens the T₁ of the myocardium and increases the signal intensity of well perfused tissues on T₁-weighted images. However, this indirect assessment of perfusion can lead to difficulties in the absolute quantification of myocardial perfusion [19]. In contrast, hyperpolarized perfusion measurements rely directly on the signal obtained from the hyperpolarized tracer and so allow for easier absolute quantification. The continual decay of the hyperpolarized signal obviously needs to be taken into account and represents a limitation of the technique, but novel acquisition schemes and quantitative models can be developed to account for these effects [20]. The application of hyperpolarized tracers for the quantitative assessment of perfusion has been demonstrated in the pig myocardium [1] and the rat brain [21].

The inherent lack of background signal when using hyperpolarized ¹³C tracers also offers potential for use in interventional MR experiments. As a demonstration of this, hyperpolarized tracers have been used to enable passive catheter tracking through the aorta of a pig [22]. Through the use of a specially modified three-lumen catheter, the flowing hyperpolarized tracer was directly imaged using a 3D back-projection imaging technique that offered frame rates of 1.4 frames/s. The obvious limitation for hyperpolarized tracers in this application is the need to continually refresh the tracer, as the signal will decay through relaxation mechanisms and the continuous RF pulsing. However, the rapid polarization offered by the PHIP technique could provide a solution to this problem.

There is a growing body of evidence to suggest that alterations in metabolic substrate utilization may be a cause, rather than a symptom, of cardiovascular disease [23‐25]. To increase our understanding of alterations in substrate selection, it is necessary to use an in vivo method that can repeatedly measure metabolism noninvasively. This would allow the metabolic changes which occur during disease progression to be assessed and would open a new window on the development of possible metabolic treatments. Magnetic resonance spectroscopy (MRS) using carbon (¹³C) is ideally suited to the study of metabolism due to the extensive range of compounds that can be detected and the ability to attribute signals to the different carbon atoms within individual molecules. ¹³C studies of cardiac metabolism have been conducted since the early 1980s when the first experiments were performed on the isolated perfused rat heart by Bailey et al. [26] using ¹³C-enriched sodium acetate. A wealth of work has since followed allowing detailed studies of metabolic substrate selection in normal [27], ischemic [28], and diabetic [29] hearts. However, the clinical utility of ¹³C MRS experiments has been limited due to a combination of the inherently low sensitivity of MRS, the low in vivo concentrations, and the low natural abundance of ¹³C. As such, the majority of cardiac studies have been performed on ex vivo hearts, perfused with ¹³C-enriched molecules and using very long scan times. This has resulted in the ability to only study metabolic steady state conditions [30] and, so far, there have only been a limited number of in vivo studies [31] or studies of non–steady state rates of oxidative metabolism [32] using non-hyperpolarized techniques.

The advent of hyperpolarization, particularly with DNP, has provided the ability to generate metabolically active “tracers” that can be injected into an in vivo experiment. The high SNR provided allows both the visualization of the injected molecule and also the downstream metabolic products to which the ¹³C label is transferred. To date, the majority of experiments with this method have focused on the key metabolic molecule pyruvate, labeled with ¹³C at the first carbon position ([1-¹³C]pyruvate). Pyruvate is metabolically important because it sits at the junction of metabolism between glycolysis and the tricarboxylic acid (TCA) cycle (Fig. 1) and therefore allows a range of different metabolic pathways to be explored. [1-¹³C]pyruvate also offers practical advantages as it is easily hyperpolarized, is rapidly taken up into cells and metabolized, and has a long relaxation time (T₁ of ~60 s) meaning that the hyperpolarized signal persists for a long time.

In the first application of DNP-generated hyperpolarized [1-¹³C]pyruvate in the perfused rat heart, Merritt et al. [10] showed the conversion of infused pyruvate into lactate, alanine, bicarbonate, and carbon dioxide. The lactate observed has since been shown to be due to rapid exchange of the ¹³C label with endogenous lactate via the enzyme lactate dehydrogenase (LDH) [33]. Similarly, the alanine signal is generated by the transamination reaction mediated by alanine aminotransferase (AAT). The bicarbonate resonance is observed due to the transformation of pyruvate into acetyl-CoA via the enzyme pyruvate dehydrogenase (PDH). During this enzymatic transfer, the ¹³C label is cleaved from the pyruvate molecule as ¹³CO₂, which is in a rapid, pH-dependant equilibrium with bicarbonate under the action of the enzyme carbonic anhydrase (CA). Therefore, the observation of bicarbonate can be seen as a surrogate marker of flux through PDH. The ability to measure flux through PDH could be vitally important as PDH is one of the key regulatory enzymes of the glucose–fatty acid (or Randle) cycle, and is therefore a key determinant of substrate selection in the heart [34].

In the first use of hyperpolarized [1-¹³C]pyruvate in the in vivo rodent heart, Schroeder et al. demonstrated that not only could the metabolic fate of [1-¹³C]pyruvate be monitored in vivo, but that physiological (overnight fasting) and pathological (type 1 diabetes) modulation of PDH flux could be observed [11••, 35]. They further demonstrated that the in vivo assessment of PDH flux in their model of type 1 diabetes correlated significantly with the severity of the disease. This work was extended by Atherton et al. [36], who demonstrated that the in vivo assessment of PDH flux obtained using hyperpolarized MRS correlated significantly with the ex vivo activity of the PDH enzyme as assessed by a traditional enzymatic activity assay. In this work a period of high fat feeding was used to reduce PDH activity, and dichloroacetate (a pyruvate analogue known to increase PDH activity) was infused to generate a range of in vivo PDH activities.

Further work has since shown the applicability of the DNP technique to the study of myocardial ischemia in the perfused heart [37, 38••], where elevated levels of lactate were observed following a period of global ischemia. Also, the ability of hyperpolarized MRI to assess intracellular pH in the in vivo heart has been demonstrated [39•]. As mentioned previously, the ¹³CO₂ produced when [1-¹³C]pyruvate is decarboxylated by PDH is in a rapid, pH-dependant equilibrium with bicarbonate (H¹³CO_3-). By measuring the relative intensities of the ¹³CO₂ and H¹³CO_3- resonances, it is possible to calculate the intracellular pH using the Henderson-Hasselbalch equation [39•].

Other metabolic pathways have been monitored in the heart. The use of [2-¹³C]pyruvate has been shown to provide a method for the real-time assessment of metabolism through the TCA cycle (Fig. 1) [38••]. The use of pyruvate that is labeled with ¹³C at the second carbon position means that the ¹³C molecule is no longer cleaved into ¹³CO₂ but is carried into acetyl-CoA instead, allowing assessment of flux into the TCA cycle. In this work, Schroeder et al. demonstrated the ability to visualize the conversion of [2-¹³C]pyruvate into the downstream metabolites citrate, glutamate, and acetylcarnitine, and further, that the flux into the TCA cycle was reduced following a period of global ischemia.

One of the key issues to consider with the application of metabolic imaging using the DNP technique is that the molecules being injected are, by their very nature, metabolically active. They are also generally injected at supraphysiological concentrations. For example, in vivo experiments with [1-¹³C]pyruvate have frequently been carried out with injections of ~80 mM pyruvate, when the normal physiological concentration is ~60 μM [36]. As such, hyperpolarized compounds generated by DNP cannot be considered as “tracers” in the strictest sense of the word and the physiological consequences of the injected compound need to be considered [36, 40, 41]. To this aim, recent work has demonstrated that, due to saturation of either the mitochondrial pyruvate transporter or PDH itself, lower concentrations of pyruvate can be injected without affecting the level of bicarbonate signal observed [41, 42••]. Atherton et al. [36] have also shown that following a 1 mL injection of 80 mM pyruvate into the rat, the maximum plasma concentration reached is only 250 μM. This relatively low concentration of pyruvate in the plasma, due most likely to the small volume injected and the rapid systemic uptake of pyruvate, is in line with pyruvate levels seen during exercise.

All of the work discussed so far on metabolic assessment with DNP has involved the use of unlocalized or “global” MR spectroscopy techniques. However, to visualize metabolic heterogeneity in the diseased heart, imaging techniques are required to distinguish healthy myocardium from diseased myocardium. As an initial approach to this, Golman et al. [9••] demonstrated the use of a single slice, chemical shift imaging (CSI) acquisition scheme to allow the spatial localization of pyruvate, lactate, alanine, and bicarbonate following an injection of [1-¹³C]pyruvate into the pig heart. They further demonstrated localized alterations in the bicarbonate image following a period of ischemia caused by temporary occlusion of the left circumflex artery. The spatial resolution in this study was 7.5 × 7.5 × 20 mm and the total acquisition time was 13.4 s (non–ECG gated acquisition). This resolution has since been surpassed by Lau et al. [42••], who demonstrated a cardiac and respiratory gated, single-shot spiral acquisition technique with a spatial resolution of 3.7 × 3.7 × 10 mm and the ability to image any specific metabolite in ~80 ms. With this technique, whole heart coverage (6 slices) can be achieved within 3 heartbeats.