Detecting treatment response in a model of human breast adenocarcinoma using hyperpolarised [1-13C]pyruvate and [1,4-13C2]fumarate

There has been substantial growth and investment in drug discovery and development programmes in oncology in recent years, with the aim of producing therapeutics targeted at the spectrum of genetic abnormalities found in cancer (Hu et al, 2000; Semenza, 2003; Bryant et al, 2005). This rapid expansion has moved the field of clinical oncology closer to the realisation of personalised medicine, in which patients receive drugs, individually or in combination, that are tailored to their specific disease. The delivery of this approach should be facilitated by noninvasive imaging methods that allow an early assessment of treatment response, identifying unsuccessful treatments at an early stage and enabling the selection of more effective treatment (reviewed in Brindle (2008a)). Currently, tumour treatment responses are assessed from reductions in tumour size (Eisenhauer et al, 2009). However, this approach lacks sensitivity and many weeks may elapse before a change in size is detected (Neves and Brindle, 2006; Weissleder and Pittet, 2008; Brindle, 2008a). In some cases, for example with cytostatic therapies, there may be no change in size despite a positive response to treatment (Brindle, 2008a). Measurements of tumour physiology or biochemistry, however, can give a much earlier indication of treatment response. Positron emission tomography (PET) measurements of the uptake of the glucose analogue [18F] 2-fluoro-2-deoxy--glucose (FDG) are being used increasingly in the clinic to stage disease and provide early evidence of treatment response (Eisenhauer et al, 2009). MRS, particularly 1H MRS, can also be used to detect the metabolic changes that accompany a positive response to treatment (Aboagye and Bhujwalla, 1999; Glunde et al, 2006). The relative lack of sensitivity of MRS, however, limits the spatial and temporal resolution of 1H spectroscopic imaging experiments and in many cases only single voxel studies are performed (Hu et al, 2009). Although 1H MRS experiments are performed widely in the clinic, this lack of sensitivity has limited their routine use. This situation may change with the recent introduction of a dynamic nuclear polarisation (DNP) technique that can increase sensitivity in the solution state 13C MRS experiment by >10 000-fold (Ardenkjaer-Larsen et al, 2003). The enormous gain in sensitivity means that, following injection of a hyperpolarised 13C-labelled cell substrate, there is sufficient signal to image the molecule in vivo, and, more importantly, its metabolic conversion into other cell metabolites. Whereas 1H MRS measurements provide a largely static picture of the levels of tissue metabolites, this labelling technique enables dynamic imaging of cellular metabolism. The principal drawback of the method, however, is the short half-life of the polarisation; for [1-13C]pyruvate this is ∼30 s in vivo, which means that the material must be injected and imaged within ∼5 min. Thus, in order to image metabolism using this technique, the hyperpolarised 13C-labelled substrate must be taken up rapidly by the cell and its subsequent metabolism must be very fast. Even then, it is often only possible to monitor a single enzyme-catalysed reaction (Gallagher et al, 2009a). Nevertheless, the technique has already shown promise for detecting treatment response in tumours (Day et al, 2007; Witney et al, 2009). The exchange of hyperpolarised 13C label between [1-13C]pyruvate and lactate, in the reaction catalysed by lactate dehydrogenase (LDH), was shown to decrease in a drug-treated murine lymphoma in vivo (Day et al, 2007), where decreased label flux was due to a number of factors, including: DNA damage-mediated activation of polyADP-ribose polymerase (PARP) and consequent depletion of the NAD(H) coenzyme pool, a loss of LDH activity, and a reduction in tumour cellularity (Day et al, 2007; Witney et al, 2009). A similar study in the same tumour model using [1,4-13C2]fumarate, showed that the rate of the fumarase catalysed conversion of fumarate to malate was a measure of subsequent drug-induced cellular necrosis (Gallagher et al, 2009b). With a clinical trial using hyperpolarised [1-13C]pyruvate about to start in prostate cancer (Brindle, 2008b), there is a reasonable expectation that the technique could translate to the clinic, where it offers a new functional imaging approach to detect early tumour responses to treatment. We show here, in a model of human breast adenocarcinoma, that a combination of hyperpolarised [1-13C]pyruvate and [1,4-13C2]fumarate can be used to detect response to doxorubicin treatment before there is any detectable change in tumour size. There was a decrease in labelled lactate production, reflecting a DNA damage response and initiation of the apoptotic program, and an increase in labelled malate production, reflecting the onset of cellular necrosis.