Stefan Markovic1, Tangi Roussel2, Keren Sasson3, Dina Preise3, Lilach Agemi3, Avigdor Scherz3, and Lucio Frydman1
1Department of Chemical and Biological Physics, Weizmann Institute of Science, Rehovot, Israel, 2Center for Magnetic Resonance in Biology and Medicine, Marseille, France, 3The Moross Integrated Cancer Research Center, Weizmann Institute of Science, Rehovot, Israel
Synopsis
Deuterium
Metabolic Imaging (DMI) was used to follow metabolism in two pancreatic cancer mouse
models, after tail-vein administration of 2H6,6’-glucose. Metabolic maps for the glucose and for its
metabolic products 2H3,3’-lactate and 2H-water
were measured over a time course of 2 h by 2H chemical shift imaging
(CSI) at 15.2 T. Abdominal images exhibited sharp and specific lactate signals,
which generated exclusively in the tumors. Thus, DMI may open valuable opportunities
for non-invasively imaging pancreatic cancer –including its diagnosis and
treatment.
INTRODUCTION
Recent studies have concluded that tracking and
imaging the fate of deuterated metabolites, could map non-invasively glycolysis
and other biochemical processes in animals and humans. A promising avenue of
this deuterium metabolic imaging (DMI) approach involves looking at externally
administered 2H6,6’-glucose, as it is taken up and
metabolized in various organs into different products as a function of time.1-6
This study employs DMI to study two strains of orthotopic pancreatic
cancer mice models (PAN-02 and KPC) at 15.2 T.
For both cases the generation of a clear 2H3,3’-lactate
signature localized solely in the tumors, evidenced the ability of this
approach to non-invasively image this manifestation of the Warburg effect.METHODS
All experiments were preapproved by an institutional
IACUC. Pancreatic cancer was investigated with two models: n=7 black mice were
implanted with PAN-02 orthotopic tumors; n=5 were implanted with KPC cells.7
All experiments were recorded in a Bruker Biospec scanner at 15.2T
running Paravision 6. A “sandwich” coil setup was used for scanning, whereby a
20x45 mm 1H butterfly surface coil (650 MHz) was placed underneath
the sidewise-laying mouse, and a customized 20 mm single-loop surface coil
tuned to 2H’s (99.8 MHz) was placed on top of the belly. Anatomical
images were collected using 1H TurboRARE (28-38 slices, 0.5 mm slice
thickness, 0.2 mm in-plane resolution). For DMI mice were catheterized in their
tail-veins, and intravenously administered 99.5%-enriched 2H6,6’-glucose
in PBS, at a dose of 3 g/kg as a single bolus 0.25 mL injection within 60 sec. Non-localized 2H
NMR data sets were acquired with a ≈20° excitation pulses, 100 ms acquisition
time, 0 ms recycle delay, 128 transients. Spatially-resolved 2H CSI
data were collected using ≈90° flip-angles (≈ 5mm slice thickness), 60 ms
acquisition times, 40x40 mm2 FOVs, 8x8 k-matrices
(zero-filled to 32x32) sampled on a Cartesian grid, and repetition times TR =
95 ms. A k-dependent weighted-average, collecting 320 signal averages for k=0 and
progressively less for its periphery, was used. This delivered a data set every
~8 min. Non-localized 2H NMR and slice-selective 2H CSI
sets were acquired in alternated, interleaved blocks spaced ~15-20 minutes,
starting from before and continuing for ca. 120 minutes after deuterated
glucose injection. 2H spectra and images were reconstructed in Matlab®
using custom written code.RESULTS
2H NMR and spatially-resolved
2H CSI data recorded after 2H6,6’-glucose
injection, showed the formation of 2H3,3’-lactate and 2H-water
for both pancreatic cancer models (Figure 1A). When implemented with
localization, these experiments only revealed 2H3,3’-lactate
in the tumor voxels (Figures 1B, 1C). Figures 2A-2C exemplify this with 2H
CSI data, where images that were spectrally-resolved for glucose, lactate and
water, are displayed for selected post-injection time points. Localization of 2H3,3’-lactate
can be detected exclusively in the tumor discernable in the 1H
anatomical image (Figure 2D); analogous behaviors were observed for other
PAN-02- and KPC-implanted mice. Typical metabolic concentration changes as a
function of time are presented in Figures 2E-2F, calculated on the basis of an
initial natural-abundance HDO background signal corresponding to 10 mM. Glucose
levels in the kidney and tumor increase with the initial bolus injection and
then gradually decrease, whilst water is steadily generated over the course of
two hours. 2H3,3’-lactate reaches lower concentrations,
but always peaks ≈60-90 min after injection. Figure 3 gives two different views of the overall metabolic
levels reached in these studies: Panel A shows them normalized by the pre-injection
water signal; panel B normalized by the maximum glucose observed for each
animal/injection. Voxels chosen for
these analyses included the kidneys as a well-defined, healthy organ, and the
tumorous tissue. The perfusion of the injected glucose is remarkably similar in
both tissues. Also similar are the lactate levels detected in the tumors, for both
pancreatic cancer strains.
The experimental time traces were fit to a simple body/tumor two-compartments
kinetic model based on the first-order differential equations shown in Figure 4A.
Here G, W and L stand for the three metabolites, the
b/t denote the compartments, and k are rates of a metabolite’s in-flow or
out- from the compartment, or for its intra-compartmental metabolic generation
from glucose. Fits of the experimental data provide two cancer-relevant rates:
the flow of glucose into the tumor, and its rate to lactate conversion inside
the tumor. Once again, no significant differences between these rates for the two
pancreatic tumor models,could be determined (Figure 4B).CONCLUSIONS
DMI’s
potential for revealing pancreatic cancer in animal models by following 2H6,6’-glucose
injections, was here demonstrated. Perfusion into the pancreatic tumors was successfully
mapped, andmetabolic transformation into water and lactateas resulting from
Krebs and glycolytic pathways followed.Acknowledgements
Support from the Israel Science Foundation, the Kimmel
and the Clore Institutes for Magnetic Resonance (Weizmann Institute), the Thompson
Family Foundation and the Israel Cancer Research Foundation, are acknowledged.References
(1)
M. Lu, X. H. Zhu, Y. Zhang, G. Mateescu, W. Chen, Quantitative assessment of
brain glucose metabolic rates using in vivo deuterium magnetic resonance
spectroscopy. Journal of cerebral blood
flow and metabolism : official journal of the International Society of Cerebral
Blood Flow and Metabolism 37,
3518-3530 (2017).
(2)
H. M. De Feyter, K. L. Behar, Z. A. Corbin, R. K. Fulbright, P. B. Brown, S.
McIntyre, T. W. Nixon, D. L. Rothman, R. A. de Graaf, Deuterium metabolic
imaging (DMI) for MRI-based 3D mapping of metabolism in vivo. Science Advances 4, eaat7314 (2018).
(3)
F. Kreis, A. J. Wright, F. Hesse, M. Fala, D.-e. Hu, K. M. Brindle, Measuring
Tumor Glycolytic Flux in Vivo by Using Fast Deuterium MRI. Radiology 294, 289-296
(2019).
(4)
M. J. Riis-Vestergaard, C. Laustsen, C. Ø. Mariager, R. F. Schulte, S. B.
Pedersen, B. Richelsen, Glucose metabolism in brown adipose tissue determined
by deuterium metabolic imaging in rats. International
Journal of Obesity 44, 1417-1427
(2020).
(5)
R. A. de Graaf, A. D. Hendriks, D. W. J. Klomp, C. Kumaragamage, D. Welting, C.
S. Arteaga de Castro, P. B. Brown, S. McIntyre, T. W. Nixon, J. J. Prompers, H.
M. De Feyter, On the magnetic field dependence of deuterium metabolic imaging. NMR in Biomedicine 33, e4235 (2020).
(6)
C. von Morze, J. A. Engelbach, T. Blazey, J. D. Quirk, G. D. Reed, J. E.
Ippolito, J. R. Garbow, Comparison of hyperpolarized 13C and
non-hyperpolarized deuterium MRI approaches for imaging cerebral glucose
metabolism at 4.7T. Magnetic Resonance in
Medicine 2020;00:1-10.
(7)
R. P. Martinho, Q. Bao, S. Markovic, D. Preise, K. Sasson, L. Agemy, A. Scherz,
L. Frydman, Indentification of variable stages in murine pancreatic tumors by a
multiparametric approach employing hyperpolarized 13C MRSI, 1H
diffusivity and 1H T1 MRI. NMR in Biomedicine, e4446
(2020).