Assessment of the sensitivity of 2H MR spectroscopy measurements of [2,3‐2H2]fumarate metabolism for detecting tumor cell death

Abstract Imaging the metabolism of [2,3‐2H2]fumarate to produce malate can be used to detect tumor cell death post‐treatment. Here, we assess the sensitivity of the technique for detecting cell death by lowering the concentration of injected [2,3‐2H2]fumarate and by varying the extent of tumor cell death through changes in drug concentration. Mice were implanted subcutaneously with human triple negative breast cancer cells (MDA‐MB‐231) and injected with 0.1, 0.3, and 0.5 g/kg [2,3‐2H2]fumarate before and after treatment with a multivalent TRAlL‐R2 agonist (MEDI3039) at 0.1, 0.4, and 0.8 mg/kg. Tumor conversion of [2,3‐2H2]fumarate to [2,3‐2H2]malate was assessed from a series of 13 spatially localized 2H MR spectra acquired over 65 min using a pulse‐acquire sequence with a 2‐ms BIR4 adiabatic excitation pulse. Tumors were then excised and stained for histopathological markers of cell death: cleaved caspase 3 (CC3) and DNA damage (terminal deoxynucleotidyl transferase dUTP nick end labeling [TUNEL]). The rate of malate production and the malate/fumarate ratio plateaued at tumor fumarate concentrations of 2 mM, which were obtained with injected [2,3‐2H2]fumarate concentrations of 0.3 g/kg and above. Tumor malate concentration and the malate/fumarate ratio increased linearly with the extent of cell death determined histologically. At an injected [2,3‐2H2]fumarate concentration of 0.3 g/kg, 20% CC3 staining corresponded to a malate concentration of 0.62 mM and a malate/fumarate ratio of 0.21. Extrapolation indicated that there would be no detectable malate at 0% CC3 staining. The use of low and nontoxic fumarate concentrations and the production of [2,3‐2H2]malate at concentrations that are within the range that can be detected clinically suggest this technique could translate to the clinic.


| INTRODUCTION
Imaging cell death can give an early indication of tumor treatment response and the effectiveness of therapy, where the degree of tumor cell death can be an indicator of treatment outcome. 1,2 Currently, assessment of tumor treatment response in the clinic focuses primarily on changes in tumor size. 3,4 However, these may take weeks to appear following treatment, reducing the therapeutic window and impacting patient survival.
Despite the introduction of several promising cell death imaging agents in preclinical studies, to date, clinical trials of these agents have failed to produce robust results. 5 We have shown previously that tumor cell death post-treatment can be detected in mouse tumor models in vivo through the production of hyperpolarized 13 C-labeled malate in animals injected intravenously with hyperpolarized [1,4-13 C 2 ]fumarate. 6,7 Fumarate is hydrated in the reaction catalyzed by the enzyme fumarase to produce malate. In necrotic cells, loss of plasma membrane integrity results in fumarate rapidly gaining access to the enzyme and an increased rate of malate production. More recently, we have shown, in EL4 murine lymphomas, in human colorectal (Colo205) and breast cancer (MDA-MB-231) xenografts, 8 and in orthotopically implanted patient-derived glioblastoma xenografts, 9 that 2 H MRS and MRSI measurements can be used similarly to detect cell death post-treatment by measuring the increase in 2 H-labeled malate concentration following intravenous [2,3-2 H 2 ]fumarate injection. Although the sensitivity of label detection is less for the 2 H experiment, the contrast, expressed as the labeled malate/fumarate ratio, is greater, as the labeled malate concentration can build up over a much longer period of time. The short half-life of the hyperpolarized 13 C label limits the degree of contrast that can be achieved in the 13

| Tumor implantation and treatment
MDA-MB-231 cells were resuspended in a 1:1 mixture of Matrigel (Corning) and complete DMEM and implanted subcutaneously on the upper back and between the fore limbs of 10-12-week-old female BALB/c Nu/Nu mice (Charles River) at 7 x 10 6 cells. 2 H spectroscopy measurements were conducted when the tumors reached $ 1 cm 3 ($35 days following implantation). The animals were treated with an escalating dose of a tumor necrosis factor (TNF)-related apoptosis-inducing ligand (TRAIL) receptor 2 agonist (MEDI3039), 10

| MR spectroscopy in vivo
Animals were anesthetized by inhalation of 2% isoflurane in air/O 2 (75%/25%, 2 L/min). Breathing rate and body temperature were monitored and body temperature was maintained with a stream of warm air. 2 H spectroscopy was performed at 7 T (Agilent, Palo Alto, CA, USA) using a home-built 10-mm diameter single-loop surface coil placed over the tumor, as described previously. 8,9,11 Tumors were localized in axial 1  and 0.2 mL was administered intravenously via a tail vein catheter to give concentrations of 0.1, 0.3, and 0.5 g/kg body weight. Infusion started 5 min after the start of spectral acquisition and lasted for a period of 20 min. Thirteen serial 5-min 2 H spectra were acquired with a 2-ms BIR4 pulse, 12 with a nominal flip angle of 67 , a TR of 140 ms and 2142 signal averages 8 over a time period of 65 min. Localization of signal to the tumors was achieved by the excitation profile of the surface coil. Spectra were zero-and first-order phase-corrected, and the peaks modeled with the AMARES algorithm implemented in the OXSA toolbox, 13,14 which uses a time-domain fitting algorithm and employs prior input knowledge to optimize the accuracy of fit. 13 The peak integrals were determined from the time domain modeling. The area under the peak was determined from the integral of the fitted line shape. The intensity of the HDO peak at the first time point was assumed to correspond to the natural abundance of deuterium in tumor water. The signals were corrected for saturation and for the number of 2 H nuclei in [2,3-2 H 2 ]fumarate, as described previously. 8 The concentration of malate was estimated from the intensity of the upfield resonance, assuming that it has a similar T 1 to the fumarate resonance.
The natural abundance of deuterium in tumor water was estimated from the measured concentration of 2 H in tap water and by assuming a tissue water content fraction of 0.8. 15 The natural abundance 2 H content of Cambridge tap water was measured by adding 5 mM formate-d, as a reference standard, to 1 mL of water, and acquiring fully relaxed 2 H spectra at 310 K using the 2 H coil of a 5-mm 1 H/broadband inverse detection probe in a 14.1-T NMR spectrometer (Bruker Spectrospin). Spectra were acquired using a 90 pulse, a TR of 10 s with a 2000 Hz spectral width, into 1024 datapoints, and were the sum of 1024 transients. Spectra were phased, baseline corrected, and the peak integrals were calculated using Topspin (Bruker Spectrospin). The amplitude of the water resonance was normalized to the formate-d peak to give an absolute 2 H concentration of 15.61 mM. The tumor HDO peak, when corrected for tissue water fraction, was therefore assumed to correspond to 12.48 mM 2 H.

| Histology and immunohistochemistry
Tumors were excised immediately following the imaging session and transferred into 10% formalin for 24 h, followed by 70% ethanol, and embedded in paraffin before sectioning into two 10-μm thick sections spaced 100 μm apart. Sections were stained with hematoxylin and eosin

| Statistical analysis
Statistical and graphical analyses were performed using Prism v. 9.0 (GraphPad). Analysis of variance (ANOVA) was used followed by Tukey's post hoc test for multiple comparisons of groups to determine significance. Pearson's R test was used to assess the significance of the correlation analysis. p values are summarized in the figures as: *p = 0.01-0.05; **p = 0.001-0.01; ***p = 0.0001-0.001; and ****p < 0.0001. Data are shown as mean ± SD, unless stated otherwise.

| RESULTS
Deuterium-labeled fumarate, malate and water concentrations were measured in subcutaneous MDA-MB-231 breast cancer xenografts following intravenous injection of [2,3-2 H 2 ]fumarate at 0.1, 0.3, and 0.5 g/kg and following treatment with 0.1, 0.4, and 0.8 mg/kg MEDI3039. The data for untreated tumors and tumors treated with 0.8 mg/kg MEDI3039 are shown in Figure 1, and the rest of the data are shown in Figure S1 in the supporting information. Individual 5-, 10-, and 20-min spectra acquired starting at 20 min following injection of 0.3 g/kg fumarate and 24 h after treatment with 0.8 mg/kg MEDI3039 are shown in Figure S2. The fumarate concentrations in the tumors at 20 min postinjection, when the concentration started to plateau, are shown in Table 1. There were significant increases in tumor fumarate concentrations with increases in injected dose and following treatment. The increase following treatment reflects an increase in tumor perfusion, as has previously been observed in this tumor model. 8 Table S1 and Figure S5. The coefficients of variance ranged from 8% to 21% for fumarate and from 6% to 35% for malate, with the values decreasing with increasing tumor metabolite concentrations.
The initial rate of malate production was estimated from the malate concentration at 20 min and plotted against the tumor fumarate concentration at this time point (Figure 2). Fitting of these data to the Michaelis-Menten equation (Equation 1) gave the values for the apparent Vmax and Km shown in Table 2.
where v is the initial rate of malate production (mM/min), [S] is the tumor fumarate concentration (mM), Km is the Km of fumarase for fumarate (mM), and Vmax (mM/min) is the maximal velocity of the enzyme at saturating fumarate concentrations. There was no increase in malate production rate at tumor fumarate concentrations greater than 2 mM, which in this tumor model was obtained with an injected fumarate concentration of 0.3 g/kg and above, and therefore the enzyme appears to be saturated at this fumarate concentration ( Figure 2). The apparent Vmax increased with increasing drug concentrations, reflecting the increased tumor cell death at the higher drug concentrations (Figure 3), and therefore the increased access of the injected fumarate to intracellular fumarase.
The extent of drug-induced cell death was determined by histological assessment of cells positive for CC3 and TUNEL (Figure 3). Increasing the drug dose resulted in parallel increases in CC3 ( Figure 3E-H,M) and TUNEL ( Figure 3I-L,N) Figure 2 to the Michaelis-Menten equation.
concentrations were linearly correlated with CC3 ( Figure 4A-F) and TUNEL staining ( Figure 4G-L) at all three injected fumarate concentrations, but both were higher at 0.3 and 0.5 g/kg.

| DISCUSSION
In a previous study with EL4 murine lymphomas and human colorectal (Colo205) xenografts, and with the breast cancer (MDA-MB-231) xenografts used in this study, we showed that following treatment there was a 10-to 18-fold increase in the tumor malate/fumarate signal ratio following intravenous injection of [2,3-2 H 2 ]fumarate at 1 g/kg. 8  Plots of malate/fumarate ratio and malate concentration versus the percentage of dead cells (CC3 and TUNEL positive) were linear with intercepts of near zero, demonstrating that there is little or no malate production in viable tumor tissue and that there is little wash in of malate from other tissues, as was demonstrated previously using blood sampling. 8 These plots allow estimation of the degree of contrast that will be generated, and therefore the sensitivity for detecting cell death, at any level of cell death for a given injected fumarate concentration. At an injected fumarate dose of 0.3 g/kg, 20% CC3 staining corresponded to a tumor malate concentration of 0.62 mM and a malate/fumarate ratio of 0.21. In principle, these plots allow estimation of the degree of cell death-dependent tumor contrast that would be generated for any tumor if it is assumed that intracellular fumarase activity and the degree of exposure of injected fumarate to the enzyme during necrotic cell death is similar between the breast cancer tumor model used here and other tumor types. In this regard, it is notable that the activities of the enzyme in EL4 murine lymphomas, human colorectal (Colo205) xenografts and breast cancer (MDA-MB-231) xenografts are similar and all three tumors showed similar cell death-dependent increases in the malate/fumarate ratio post treatment. 8 In orthotopically implanted patient-derived xenograft models of glioblastoma treated with chemoradiation, we observed even greater contrast with 20% CC3 staining, corresponding to a malate concentration of 1.38 and a malate/fumarate ratio of 0.71. 9 In a study in breast cancer patients treated with chemotherapy, biopsy sampling showed a median increase of 3.4% of TUNEL positive cells in responders compared with À0.1% in nonresponders. 18 However, there was a marked variation in TUNEL staining, which may reflect inadequate sampling of the tumor mass by biopsy, with some responding patients showing large increases in staining, from 0.6% to 6.2%, 8.8% to 15%, 1.7% to 8.2%, and 0.4% to 1.6%. Similar results were obtained in a more recent study in patients treated with trastuzumab, where there was a median increase in CC3 staining, from 3.5% to 4.7%, with again a wide variation in the increases between patients, with some showing an apoptotic index of greater than 10% at 1 week after treatment. 19 Assuming cell death doubles from 5% to 10% post-treatment, then based on the data presented here for MDA-MB-231 xenografts, injection of 0.3 g/kg fumarate would give a 40% increase in the malate/fumarate ratio. Moreover, if the image voxel encompasses the entire volume of the tumor then the experiment effectively integrates all the cell death that is present and thus avoids the sampling error that plagues biopsy.
An advantage of the technique is that it probably only detects necrotic cells that have been killed recently by treatment because fumarase is likely to leak out of treated tumors over time. Cells that were killed by previous treatments are less likely to be detected. A potential drawback of the technique is that it is less likely to detect cell death if the number of necrotic cells increases slowly or if the timing of the increase in cell death post-treatment is unknown. The treatments that we have used here and in previous studies with hyperpolarized 13 C-labeled 6,7 and 2 H-labeled fumarate 8,9,11 led to relatively rapid increases in tumor cell death.  11 If we assume that, in general, an oral dose must be twice that of an injected dose, then for a 60-kg patient an injected dose of 0.3 g/kg would be equivalent to an oral dose of 36 g [2,3-2 H 2 ] fumarate. Fumarate has been given orally to humans in doses ranging from 5 to 30 g to study its potential laxative effects. Up to six doses were given with no evidence of renal or liver impairment, 22 demonstrating the feasibility of using oral [2,3-2 H 2 ]fumarate and 2 H MRSI to image tumor cell death in the clinic.

| CONCLUSION
We showed previously that tumor cell death could be detected from the labeled malate produced following intravenous injection of 1 g/kg [2,3-2 H 2 ]fumarate. We have shown here that cell death can still be detected with injected [2,3-2 H 2 ]fumarate concentrations as low as 0.1 g/kg, although 0.3 g/kg was optimal, and that the malate/fumarate ratio was linearly dependent on the extent of tumor cell death.