Deuterium MR Spectroscopy: A New Way to Image Glycolytic Flux Rates

See also the article by Kreis et al in this issue.

Dr Ronald Ouwerkerk is a biochemist and expert on MRI and MR spectroscopy with nuclei other than ¹H. He worked on ³¹P MR spectroscopy at University of Oxford (Oxford, England) until 1992, then at Utrecht University and at Philips Medical Systems (Best, the Netherlands). He was recruited in 1998 to work at Johns Hopkins University (Baltimore, Md) on ³¹P and ²³Na MRI. In 2008, he joined the National Institute of Diabetes and Digestive and Kidney Diseases (Bethesda, Md), where he works as scientific staff on metabolic imaging.

There are many imaging modalities that can provide information on metabolism. MRI by itself provides a multitude of these tools. The most widely used MRI tool is MR spectroscopy. MR spectroscopy can provide detailed localized information about metabolism in vivo. However, unless it is used with some sort of direct or indirect metabolic challenge, MR spectroscopy only yields information on the steady state concentration of sufficiently abundant metabolites. Other MR spectroscopic techniques can yield information about metabolite fluxes, such as phosphorus 31 (³¹P) magnetization transfer or carbon 13 (¹³C) with administration of ¹³C-enriched substrates, but these require specialized equipment, are technically challenging, and typically do not yield high-spatial-resolution images.

Experiments with hyperpolarized ¹³C substrates such as pyruvate can yield metabolic images with very good resolution and important metabolic flux information (1). Though technically challenging, this method is being developed for clinical use. Molecular probes for hyperpolarized ¹³C are limited to species with a long T1. Pyruvate as a probe is useful in revealing the flux to mitochondrial metabolism and the redox state, but because this probe is introduced at the end of glycolysis, this technique does not provide information on the upstream part of the glycolytic flux or glucose uptake.

PET is also a great tool for studying metabolism, but the most commonly used tracer, fluorine 18 deoxyglucose, serves specifically to trace glucose uptake. Glucose uptake is tightly regulated by the glycolytic flux and thus, indirectly, a good indicator of glycolytic activity. However, the tracer molecule is not metabolized further than 2-deoxyglucose-6-phosphate and thus inhibits glycolytic pathway after the initial step. Any steps beyond phosphorylated glucose formation are not revealed with this method.

Deuterated compounds can also be observed with MRI and they are somewhat less expensive than ¹³C-labeled compounds. Like ¹³C-labeled probes, they are metabolically equivalent to the unlabeled substrates and not radioactive. By gyromagnetic ratio and at 100% abundance, the sensitivity of deuterium is comparable to that of ¹³C, but the T1 is shorter. The long T1 of many ¹³C metabolite resonances (for pyruvate > 60 seconds) is an advantage for use in hyperpolarized ¹³C MRI, but a hindrance when using ¹³C MRI without hyperpolarization. The shorter T1 of deuterated compounds of around 300 msec for both water and lactate and about 65 msec for glucose allows for more efficient MR spectroscopic imaging sequences with less saturation effects.

Deuterium (hydrogen 2 [²H]) MR spectroscopic imaging has been demonstrated to show metabolic differences in glycolysis of tumors versus healthy tissue in rats and in humans (2). It was shown that it was feasible to quantify the deuterated products of glycolysis by using glucose with two deuterium labels on the 6-position carbon (6,6’-²H-glucose).

This is taken a step further in the study by Kreis et al (3) in this issue of Radiology by creating quantitative images of glycolytic flux in a murine model of lymphoma. In the first step, the authors tested a numerical model for measuring glycolytic flux that had been previously used for ¹³C-labeled experiments. They collected a series of surface coil–localized spectra with a time resolution of just over 1 minute. MR spectroscopy of blood samples and tumor cell cultures were also performed to validate the model and determine some of the input parameters and the minimum time resolution. A ²H MR spectroscopic imaging sequence with adiabatic excitation pulses was optimized and reconstructed with a denoising technique to achieve the required temporal resolution of 10 minutes. This allowed the authors to observe early effects on the glycolytic flux 48 hours after etoposide treatment. The resulting images mapping the glycolytic flux show heterogeneity in the glycolytic flux within the tumor before and after treatment.

This is a useful technique for preclinical studies on animal models, but deuterium imaging has also been shown to be translatable to human scale with 4.0-T field strength (2). The prospects for similar studies to be performed in humans look good. The authors used a weighted k-space sampling where the number of averages was varied as a function of k-space radius rather than the circular k-space imaging used by De Feyter et al (2). Combined with the denoising algorithm, the spatial and temporal resolutions were sufficient to achieve these dynamic results in a mouse at 9.4 T. There are multiple sparse and weighted sampling schemes and reconstruction techniques available to allow this to be translated to human studies at 7.0 T and possibly even 3.0 T.

The tumor model used for this study is an ideal candidate because its metabolism relies heavily on glycolysis. However, the treatment agent etoposide does not primarily act on glycolysis but on DNA replication. Thus, the effects on glycolysis shown in this study are likely the result of cell death. The methods used in this study would be ideal to examine other agents that more directly target the glycolytic flux. When studying other, less uniquely glycolytic tumor models, or noncancerous tissues, the deuterated water (HDO) formation and products of oxidative phosphorylation become important. Other models may have to be used to create metabolic flux maps, but even changes in absolute concentrations in ²H-labeled metabolites or ratios of labeled metabolites may be good markers of disease or therapeutic effects.

Clearly, these initial studies to using deuterium MR spectroscopic imaging are just the beginning of an exciting direction of preclinical and clinical cancer studies.