極T代謝磁共振全球科研集錦

195

0278-0062 (c) 2019 IEEE. Personal use is permitted, but republication/redistribution requires IEEE permission. See http://www.ieee.org/publications_standards/publications/rights/index.html for more information.

This article has been accepted for publication in a future issue of this journal, but has not been fully edited. Content may change prior to final publication. Citation information: DOI 10.1109/TMI.2019.2926437, IEEE

Transactions on Medical Imaging

Abstract— Kinetic modeling of the in vivo pyruvate-to-lactate

conversion is crucial to investigating aberrant cancer metabolism

that demonstrates Warburg effect modifications. Non-invasive

detection of alterations to metabolic flux might offer prognostic

value and improve the monitoring of response to treatment. In

this clinical research project, hyperpolarized [1-

13C] pyruvate

was intravenously injected in 10 brain tumor patients to measure

its rate of conversion to lactate (kPL) and bicarbonate (kPB) via

echo-planar imaging. Our aim was to investigate new methods to

provide kPL and kPB maps with whole-brain coverage. The

approach was data-driven and addressed two main issues:

selecting the optimal model for fitting our data and determining

an appropriate goodness-of-fit metric. The statistical analysis

suggested that an input-less model had the best agreement with

the data. It was also found that selecting voxels based on postfitting error criteria provided improved precision and wider

spatial coverage compared to using signal-to-noise cutoffs alone.

Index Terms— Brain Cancer, Dissolution Dynamic Nuclear

Polarization, Hyperpolarized MRI, Kinetic Modeling, kPL, kPB,

Metabolic Imaging.

I. INTRODUCTION

issolution Dynamic Nuclear Polarization (dDNP)1,2 is a

powerful technique that enhances nuclear polarization by

up to 5 orders of magnitude.

Its chemical3–5

, biological6–9, physical10–14, pre-clinical15–21

and clinical22–28 potentials have been investigated in the last

two decades. The use of hyperpolarized (HP) substrates has

provided important insights into cancer metabolism6,22,29–31

and cardiac imaging23,32–36. Thus far, HP pyruvate is the most

common substrate for in-vivo applications of dDNP because of

both its relatively long T1(

13C), high biological relevance, and

rapid uptake and conversion.

The first phase I clinical trial of HP [1-

13C]pyruvate was

reported in 201322 on patients with prostate cancer and it

1

Dep. of Radiology and Biomedical Imaging, University of California San

Francisco, San Francisco, CA, USA. 2

Dep. of Neurological Surgery, University of California San Francisco,

San Francisco, CA, USA.

We thank the funding sources: NICO and NIH grants P01CA118816,

R01EB017449 and P41EB013598. As for the first author, the views expressed

are purely those of the writer and may not in any circumstances be regarded

as stating an official position of the European Commission.

Corresponding author: daniele.mammoli@uscf.edu

demonstrated feasibility and safety for this approach to

monitoring cancer progression and treatment response by

quantifying the Warburg effect37 that is expressed through

increased lactate dehydrogenase (LDH) activity. Several

institutions around the globe are now performing clinical

cancer research of HP pyruvate in prostate, breast38, liver39

and brain28, in order to investigate the clinical value of dDNP

HP MRI and its advantages over current molecular imaging

methods including Fludeoxyglucose Positron Emission

Tomography (FDG-PET)40,41.

FDG-PET is an imaging technique used for the diagnosis of

cancer based on the uptake and trapping of the radioactive

substrate, but it does not track its kinetic conversion into

downstream metabolites. Furthermore, it is of limited use in

brain tumors since high glucose uptake is observed not only in

cancerous tissues, but also in healthy cortical gray matter42,43,

confounding the discrimination of brain tumors. There is an

unmet clinical need for a more precise diagnosis of brain

tumors which might be overcome by the capability of dDNP

to track the pyruvate-to-lactate conversion and the Warburg

effect, provided that a robust and reproducible kinetic

modeling of the conversion of pyruvate is achieved44–51.

The goal of this work is to develop, implement and test

methods for calculating precise kPL and kPB maps describing

the enzymatic conversion of [1-

13C]pyruvate to lactate and

bicarbonate in the human brain. We acquired 22 datasets

though a dynamic 2D multislice Echo Planar Imaging (EPI)52

sequence. Dynamic acquisition of MR images has the

advantage to provide robust quantification of kinetic

processes53, regardless of differences in bolus delivery, which

influences, for instance, area-under-curve (AUC) ratios. EPI

acquisition of HP substrates offers improved temporal

resolution and greater spatial coverage than Echo Planar

Spectroscopic Imaging (EPSI) techniques. We present and

compare several models and two approaches to cut off voxels

with improper fitting, in order to provide precise and spatiallyresolved kinetic maps.

Kinetic Modeling of Hyperpolarized Carbon-13

Pyruvate Metabolism in the Human Brain

Daniele Mammoli1, Jeremy Gordon1, Adam Autry1, Peder E. Z. Larson1, Yan Li1, Hsin-Yu Chen1,

Brian Chung1, Peter Shin1, Mark Van Criekinge1, Lucas Carvajal1, James B. Slater1, Robert Bok1,

Jason Crane1, Duan Xu1, Susan Chang2 and Daniel B. Vigneron1.

D

極T代謝磁共振全球科研集錦

196

0278-0062 (c) 2019 IEEE. Personal use is permitted, but republication/redistribution requires IEEE permission. See http://www.ieee.org/publications_standards/publications/rights/index.html for more information.

This article has been accepted for publication in a future issue of this journal, but has not been fully edited. Content may change prior to final publication. Citation information: DOI 10.1109/TMI.2019.2926437, IEEE

Transactions on Medical Imaging

II. METHODS

Sample preparation

The sample preparation was carried out in collaboration with

UCSF Radiopharmacy. Each sample consisted of a mixture of

Good Manufacturing Practices (GMP) [1-

13C]pyruvic acid

(14.2 M; MilliporeSigma Isotec) and 15 mM Electron

Paramagnetic Agent (EPA), (AH111501, GE Healthcare).

Samples were prepared the same morning of each study and

they were inserted, under aseptic conditions, into pharmacy

kits from GE Healthcare.

Polarization

A SPINlab54 polarizer (GE Healthcare) was used to polarize

[1-

13C]pyruvate at 5 T and about 0.8 K. The sample was

irradiated with microwave at fMW = 140.02 GHz for at least 2

hours, boosting the 13C polarization up to 40% in the liquid

state back-calculated to the time of dissolution.

Pre-injection quality check

After the dissolution, the EPA was filtered mechanically and

quality control was carried out prior to injection: pH, pyruvate

and EPA concentrations, polarization, and temperature were

measured and compared to ranges of acceptability.

Contemporarily, the integrity of the 0.2 μm sterile filter was

tested following manufacturer specifications on its resistance

to a pressure-stress of 50 psi. Following the release by the

pharmacist, ~250 mM pyruvate was injected at a dose of 0.43

mL/kg and a rate of 5 mL/s, followed by a 20 mL saline flush.

A pause of 5 s preceded the acquisition.

Acquisition

All experiments were performed on a 3T clinical MR scanner

(MR750, GE Healthcare) with multi-nuclear excitation and

32-channel reception capability. Carbon-13 RF excitation was

achieved with a volume transmit coil and detected with

custom 13C RF coils with either 8 or 32 channels55.

Images of pyruvate, lactate and bicarbonate were acquired

through a 2D EPI dynamic sequence53, capable of exciting

multiple 13C frequencies sequentially, and having the

following characteristics: 8 slices, 1.5 or 2 cm thick axial slice,

TR = 62.5 ms, TE = 21.9 ms, 20 time-points with a 3 s

temporal resolution, in-plane resolutions of 1.2 x 1.2, 1.5 x 1.5

or 2 x 2 cm2

. In order to improve image quality, we applied

noise pre-whitening56 in k-space prior to sum-of-squares coil

combination.

In Fig. 1 we show the temporal trends of the signals of

pyruvate, lactate and bicarbonate in a single slice.

III. KINETIC MODELING

In the last decade, substantial efforts have been dedicated to

developing kinetic models of MRI signals of HP metabolites

in vivo. Despite several approaches available in literature 14,25–

30, no gold-standard is available because strategies may vary

according to specific experimental conditions.

Since kinetic modelling might suffer from errors induced by

several factors (e.g., variable bolus delivery, acquisition

sequence, magnitude images with non-zero mean noise etc),

discarding voxels with poor fitting is critical to provide precise

kinetic maps.

In this section, we describe kinetic models and fitting criteria

to address these issues and to quantify kinetic rate constants in

a precise way.

First, suitable models for fitting our data are presented.

Then, we describe two criteria to decide how to discard voxels

with suboptimal fitting.

Models

While studying the human brain, in order to obtain insights

into complicated metabolic pathways, it is crucial to

investigate the kinetics of several metabolites in different

cellular environments. Although it is an oversimplification, we

consider a three-site system (i.e., pyruvate, lactate and

bicarbonate) in a single compartment. In such a system, the

magnetization ^n)o is exchanged and relaxes according to the

following equations (single macrons refer to vectors, double

macrons to matrixes):

@

@L ^n)o = _ ? ^n)o ` ]n)o [1]

_ = h

a 0a&98a&95 &89 &59

&98 a 0 a &89 0

&95 0 a 0 a &59

k [2]

^n)o = r

:;<n)o

=>?n)o

>C?n)o

s [3]

]n)o = r

n)o

n)o ? "=>?

n)o ? ">C? s *%#(# n)o = , ? #PL ? )Q [4]

We defined:

 &98 and &95 as the rates of labelling exchange from

pyruvate to lactate and to bicarbonate (respectively).

Note they are first order apparent rate constants and

not absolute flux measurements.

 R1 = 1/T1 as spin longitudinal relaxation rate.

Figure 1. Temporal dynamics of the signals of pyruvate, lactate and bicarbonate in a single slice (data set 3 in Table 1), acquired with the EPI dynamic

sequence (see text for details). For anatomic reference, a 1H FLAIR image on the left. Note that data have not been coil corrected and reflect the sensitivity

profile of the 8-channel receive-array used in this study.

極T代謝磁共振全球科研集錦

197

0278-0062 (c) 2019 IEEE. Personal use is permitted, but republication/redistribution requires IEEE permission. See http://www.ieee.org/publications_standards/publications/rights/index.html for more information.

This article has been accepted for publication in a future issue of this journal, but has not been fully edited. Content may change prior to final publication. Citation information: DOI 10.1109/TMI.2019.2926437, IEEE

Transactions on Medical Imaging

 6<9

:;< n)o as the experimental temporal trend of the

magnetization of pyruvate.

 ^6<9n0o as the experimental magnetization vector

^n)o at the first time point.

 ^6<9no as the experimental magnetization vector

^n)o at the last time point.

Overall, 5 approaches were tested:

A) n)o = 0, &89 = &59 = 0, :;<n)o = 6<9

:;< n)o,

^n0o = ^6<9n0o, ^no = ^6<9no.

This is a model with no input-function57: the fitted

signal of pyruvate is bounded to the experimentally

acquired signal. The first and last time points are

fixed to the experimental values (for all metabolites).

Just forward reactions are considered.

Fitted parameters: 0, &98, &95.

B) Same as A) except for having now &89 d 0 (back

reaction now allowed with a different rate). Fitted

parameters: 0, &98, &95,&89.

C) Same as A) except for having now ^;no =

3

BCL [1 1 1] (last time points are now estimated

from the fitting). Fitted parameters:

0, &98, &95,3

BCL .

D) Same as A) except for having now ^;no =

3

BCL [1 1 1] and ^;n0o =

/

BCL [1 1 1] (both first and last time points are

estimated from the fitting). Fitted parameters:

0, &98, &95,3

BCL , /BCL .

E) &89 = &59 = 0, ^n0o = ^6<9n0o, ^no =

^6<9no. In this model n)o d 0, meaning that there

is an input function n)o (defined in Eq. 4) simulating

the arrival of the bolus in the region of interest. The

magnetization of pyruvate is also fitted. First and last

points are set equal to experimental values. Just

forward reactions are considered. Fitted parameters:

0, &98, &95,,, -, ., "=>?, ">C?.

The differential equations 1-4 were solved through a discrete

model. The transverse magnetization, ^Z;, and the residual

longitudinal component along the z axis, ^<[K, were computed

in Eq 5 and 6 at each time step ) b  (where ) = TR and

N = 1, 2, …, 20), taking into account the excitation flip angles

,] for each metabolite.

The magnetization at time step N+1 was calculated in Eq. 7 by

considering the exchange matrix _, ) and ^<[Kn) b o.

^<[Kn) b o = ^n) b o b 

n,]o [5]

^Z;n) b o = ^n) b o b

 n,]o [6]

^n) b [ ` 1]o = #7_bL^<[Kn) b o [7]

All models were solved via custom MATLAB routines

involving minimization of non-linear least-square problems

(i.e., difference between fitted and experimental

magnetization). skPL is half the 95% confidence interval,

calculated with MATLAB functions lsqnonlin and nlparci. In

all cases, boundary conditions were: 1/80 s-1 < R1 < 1 s-1 and

0 < kPL < 1 s-1

.

Experimental T1s of lactate and pyruvate vary significantly in

solution. However, a recent work53 has shown that the inputless model is insensitive to errors in pyruvate T1 and that T1

values for pyruvate and lactate are similar in vivo (respectively

30 and 25 s, in patient studies at 3 T). We, therefore, chose to

have equal T1 values to improve the stability of the model.

After fitting to experimental data, models were classified

according to 3 indicators, Akaike Information Criterion (AIC),

ghZ

:[<? and &]

98

:[<?, defined as:

 = 2& a 2 no [8]

ghZ

:[<? = 100 GZW[

T

GZW[

S [9]

&]

98

:[<? = 100 g1 a l

&]98

C

&]98

e 4 mj [10]

where k is the numbers of parameters estimated in the model,

L is the maximum value of the likelihood function for the

model (i.e., chi squared in our case), ghZ

C is the number of

voxels, in all data sets, that are fitted according to the i

th model

(with i = A-E), ghZ

4 is the ghZC corresponding to model A,

&]98

C is the mean kPL, averaged over all data sets, for the i

th

model and &]98

4 is the &]98C corresponding to model A.

The AIC is a measure of the statistical significance of several

models with a different number of parameters to be estimated

(i.e., the lower the better statistical significance for the model).

ghZ

:[<? is informative of how many voxels are fitted if

compared to model A (chosen arbitrarily as a standard) and

&]

98

:[<? is a measure of how much kPL varies with models.

To assess the reproducibility of the pyruvate bolus delivery,

we define a “mean pyruvate time” )><<

{[>G as the center of mass

of the pyruvate signal over time on a voxel-wise basis53:

)><<

{[>G =  :X\\Y V nGoLnGo

 :X\\Y V nGo [11]

Table 1. Summary of the experimental conditions of the 22 human brain data

sets. We report coil used, voxel size, flip angles of excitation for each

metabolite, peak SNR (along all voxels and slices in the given data set), mean

kinetic rate kPL and its standard deviation (calculated with “model A” and the

error criterion, described later on in the manuscript) and mean and standard

deviation of the mean arrival time of pyruvate tarr.

極T代謝磁共振全球科研集錦

198

0278-0062 (c) 2019 IEEE. Personal use is permitted, but republication/redistribution requires IEEE permission. See http://www.ieee.org/publications_standards/publications/rights/index.html for more information.

This article has been accepted for publication in a future issue of this journal, but has not been fully edited. Content may change prior to final publication. Citation information: DOI 10.1109/TMI.2019.2926437, IEEE

Transactions on Medical Imaging

Table 2. Comparison of models A-E according to Akaike Information

Criterion (AIC) and percentage deviation from model A of mean kPL (&]

98

:[<?)

and of number total voxel fitted (ghZ

:[<?).

Criteria to mask voxels on kPL maps

In this paragraph, we introduce two approaches to mask

voxels with improper fitting in order to provide “reasonable”

kPL maps with an acceptable compromise between precision of

the fitting in a voxel and the numbers of voxel for which the

fitting itself is retained (i.e. the coverage). Due to low SNR,

kPB maps describing the conversion of pyruvate into

bicarbonate will not be considered in this analysis. In the

"SNR criterion", prior to fitting temporal trends to a model, we

create a mask discarding voxels with peak SNR lower than a

set of two SNR thresholds, one for pyruvate and one for

lactate, labeled as ?SNR:;<

{>Z, SNR=>?

{>Zü. In the "error criterion",

prior to fitting temporal trends to a model, we mask voxels

based on ?SNR:;<

{>Z, SNR=>?

{>Zü, as in the SNR criterion, but,

after fitting, we discard all the voxels where skPL is bigger

than a fraction of kPL itself.

In this way, the global masking procedure depends on both the

set of SNRs and on the statistical outcome of the fitting.

To compare the outcomes of fitting with the two criteria, we

will take into account ghZ

:[<? (as in Eq. 9 but now the index

refers to the criterion) and R2

, defined as:

1 = 1 a  n;U2BUoR U

 n;U2;]oR U

[11]

where yk and fk refer to the experimental and fitted signals at

the kth time point and +] is the average over time of yk. R2 is

preferred over AIC or chi squared since these last two vary

with different SNR (i.e., if data with lower SNR are filtered,

the residuals n+D a $Do1 are bigger).

Figure 2. [Top] 1H FLAIR and 13C total signal AUC images (data set 3) from

the raw data. [Bottom] Temporal dynamics of signals of pyruvate, lactate and

bicarbonate in the selected voxel.

IV. RESULTS AND DISCUSSION

The subject population comprised 10 patients with recurrent

glioma who had received prior surgery and were undergoing

standard-of-care treatment, including radiation therapy and

temozolomide chemotherapy, along with a variety of adjuvant

therapies. A subset of these patients was imaged serially, and

all received a single injection per day. Table 1 includes the

experimental conditions under which the 22 data sets were

acquired (i.e., coil, voxel size, excitation flip angles) and

estimation of SNRmax (the peak SNR in all voxels and slices in

the given data set), )><<

{[>G (mean arrival time of pyruvate) and

&98

{[>G (mean kPL rate).

In Fig. 2, we can see typical temporal trends of the signal-tonoise ratio of pyruvate, lactate and bicarbonate in a single

voxel, selected from a slice (also shown for all metabolites).

The SNR did not go to zero at later time-points due to the

sum-of-squares coil combination used in this work, which

results in non-zero mean noise caused by noise rectification of

magnitude data.

While there has been no comprehensive study on the impact of

spatial resolution on the measured reaction rate, larger voxels

would have a greater risk of partial-volume effects, reducing

the measured reaction rate because of the strong vascular

pyruvate signal58. Confounding factors such as Rician bias in

magnitude data and variations in bolus delivery between

patients are global factors that impact quantification. On a

voxelwise level, regional variations in perfusion, signal loss

from susceptibility/B0 inhomogeneities and reduced SNR in

voxels at the center of the brain, farthest from the

multichannel receive array, can also impact quantification.

The choice of flip angles will also have an impact on the

precision of rate constant estimates57,59,60. Using a variable

(across frequency) flip angle scheme, with a higher flip angle

for downstream metabolites and a lower flip angle for

pyruvate, is commonly used to compensate for the lower

metabolite concentration and to reduce saturation of pyruvate.

This leads to more precise fitting in the presence of noise by

increasing the metabolite SNR61. While variable (through

time) flip angle schemes can improve SNR, they are also more

sensitive to transmit B1 inhomogeneity59,62.

The SNRmax was excellent for pyruvate and lactate (i.e.,

greater than 100 and 13, respectively, for all datasets) but

greatly variable for bicarbonate.

The mean arrival time of the pyruvate bolus, )><<

{[>G, was

calculated for every data set via Eq. 11. While there is

variability throughout the brain due to differences in arrival

time between arterial and venous voxels, the mean pyruvate

time ranged from 6 s to 11 s post injection, with an average

value of 8 ± 2 s across patients. Compared to initial results

from the prostate63, the mean arrival time in the brain appears

fairly reproducible in our initial cohort.

Mean kPL and its standard deviation were also calculated on

each data set, with the approach described in the next

paragraphs.

Let us define skPL as half the 95% confidence interval from

non-linear least-square fitting to “model A” with the error

criterion with ?SNR:;<

{>Z, SNR=>?

{>Zü = [20, 5] and skPL < 0.6·kPL.

極T代謝磁共振全球科研集錦

199

0278-0062 (c) 2019 IEEE. Personal use is permitted, but republication/redistribution requires IEEE permission. See http://www.ieee.org/publications_standards/publications/rights/index.html for more information.

This article has been accepted for publication in a future issue of this journal, but has not been fully edited. Content may change prior to final publication. Citation information: DOI 10.1109/TMI.2019.2926437, IEEE

Transactions on Medical Imaging

We have applied the error criterion with ?SNR:;<

{>Z, SNR=>?

{>Zü

= [20, 5] and skPL < 0.6·kPL to compare models A to E. Values

of AIC, &]

98

:[<? and ghZ

:[<? are reported in Table 2: model A

was the best model at describing data (highest ghZ

:[<? and

lowest AIC) and model B the worst (high AIC and negligible

ghZ

:[<?). Model E, which is the only one including an input

function, performed worse than model A (the simplest

assumption-less model), due to a poor agreement between the

input function and our data which may be improved with an

earlier start of acquisition.

Parameters used in both the SNR and the error criterion

should be optimized in order to find a compromise between

precision and number of voxels fulfilling the statistical

requirements based on skPL (i.e., spatial coverage).

The analysis has been carried out on datasets with peak SNR

of lactate lower than 25, by using model A, since it is the one

providing the best fit quality.

In the SNR criterion, we have to choose ?SNR:;<

{>Z, SNR=>?

{>Zü,

namely the set of thresholds for masking voxels with low SNR

(before fitting). We tested 22 sets obtained by all the possible

combinations between SNR:;<

{>Z = [5, 10, 50, 100] and

SNR=>?

{>Z = [2, 3, , 5, , 7].

In the error criterion, the SNR mask was computed based on

6 threshold sets, obtained by combining SNR:;<

{>Z = 5 and

SNR=>?

{>Z = [2, 3, , 5, , 7]. After fitting, we discarded all the

voxels with skPL < c?kPL where c = [0.4, 0.6, 0.8, 1.0].

Table 3a. Values of ghZ

:[<? for the SNR criterion and the error criterion using

model A as standard. Several combinations of ! :;<

{>Z and ! =>?

{>Z were

used for both criteria. In the SNR criterion no conditions are set on skPL

(error of kPL), whereas, in the error criterion, fitting is accepted for skPL <

c·kPL and rejected otherwise.

Table 3b. Values of R2 for the SNR criterion and the error criterion using

model A as standard. Several combinations of SNRpyr and SNRlac were used

for both criteria. In the SNR criterion no conditions are set on skPL (error of

kPL), whereas, in the error criterion, fitting is accepted for skPL < c·kPL and

rejected otherwise.

Values of ghZ

:[<? and R2 are reported in Tables 3a and 3b

(respectively) for both the SNR criterion and the error

criterion. To produce kinetic maps, it is necessary to find a

compromise between statistical precision and available

number of voxels (or alternatively the voxel count within the

FOV) and the final decision may vary from one user to the

other.

Tables 3a and 3b provide insight on how to set limits in our

case. In Tab. 3a, ghZ

:[<? drops with higher SNR:;<

{>Z, higher

SNR=>?

{>Z and lower c, showing that any stricter requirement

diminishes the numbers of accepted voxels. For the SNR

criterion (Tab. 3b), R2 increases with higher SNR=>?

{>Z but not

necessarily with higher SNR:;<

{>Z. This suggests that the

precision of methods based on the SNR alone depends more

critically on lactate rather than on pyruvate. In Tab. 3b, as for

the error criterion, R2 increases with both higher SNR=>?

{>Z and

lower c, meaning that both parameters affect the statistical

precision.

In Tab. 3b, let us consider the R2 values corresponding to

SNR:;<

{>Z = 5 in both criteria (i.e., first column in the SNR

criterion and all columns in the error criterion). Values of R2

in case of SNR=>?

{>Z = 2, 3,  in the SNR criterion are smaller

than any R2 in the error criterion (regardless of c). However,

R2 for SNR=>?

{>Z = 5, , 7 in the SNR criterion are comparable

to those in the error criterion. This means that at low SNRlac

the error criterion performs better than the SNR criterion,

while they are comparable at higher SNRlac.

By comparing Tab. 3a and 3b, we can see that, at similar

precisions, the error criterion gives, on average, more voxels

with proper fitting (i.e., higher ghZ

:[<?), yielding to a better

spatial coverage in kPL maps. The trend is particularly evident

at low SNR=>?

{>Z.

Figure 3. Temporal dynamics of the signal of lactate in a voxel (blue crosses)

and fitting (red line) using model A and the error criterion in 6 cases with

different values of the parameter c in the condition skPL < ckPL (see text for

details).

極T代謝磁共振全球科研集錦

200

0278-0062 (c) 2019 IEEE. Personal use is permitted, but republication/redistribution requires IEEE permission. See http://www.ieee.org/publications_standards/publications/rights/index.html for more information.

This article has been accepted for publication in a future issue of this journal, but has not been fully edited. Content may change prior to final publication. Citation information: DOI 10.1109/TMI.2019.2926437, IEEE

Transactions on Medical Imaging

For example, let us consider two cases with R2 = 0.43 in Tab.

3b: those with ?SNR:;<

{>Z, SNR=>?

{>Zü = [100, 2] for the SNR

criterion and ?SNR:;<

{>Z, SNR=>?

{>Zü = [5, 3] with skPL < 1.0?kPL

for the error criterion. These two cases exhibit the most

dramatic difference since their corresponding values of ghZ

:[<?

in Tab. 3a would be 6 and 33, meaning that the error criterion

provided 5 times more voxels despite having similar statistical

precision.

Furthermore, if one wanted to base the selection on the SNR

alone, removing the check on skPL, in order to choose precise

values, it would be necessary to perform many fittings at

different SNR thresholds in every dataset and to fix a

reasonable threshold.

The error criterion is an easier option since it is sufficient to

set SNR:;<

{>Z and SNR=>?

{>Z to reasonably low thresholds (high

enough to avoid to retain voxels with noisy trends) and leave

the main filtering task to the criterion based on skPL.

We investigated the performances of the error criterion when

applied to random noise.

To do so, we generated several sets of lactate signals

consisting of Gaussian noise and associated them to the

experimental signals of pyruvate from data set 3. Then, we

applied the error criterion with ?SNR:;<

{>Z, SNR=>?

{>Zü = [5, 3]

and skPL < 0.6?kPL and calculated the number of fitted voxels,

namely those where the fitting fulfils the requirements of the

error criterion. We found that, if compared to experimental

signals from data set 3, random noise provides, at best, 40

times less fitted voxels.

This suggests that the improvements introduced by the error

criterion on ghZ

:[<? and R2 in Tables 3a and 3b are barely

influenced by the noise.

All these findings speak in favor of the error criterion, which

seems to be more reasonable than the SNR approach at

masking voxel to be displayed in kPL maps.

The choice of ?SNR:;<

{>Z, SNR=>?

{>Zü and c is arbitrary and

looking at the outcome alone might be misleading, as shown

in Fig. 3. On the quest to opt for specific values, we started by

considering the approach which has been often used in our

group so far, namely ?SNR:;<

{>Z, SNR=>?

{>Zü = [10, 5]. According

to tables 3a and 3b, it has ghZ

:[<? = 17 and R2 = 0.53.

Bearing this in mind, we created kPL and kPB64 maps for every

data set, by using model A combined with the error criterion

with ?SNR:;<

{>Z, SNR=>?

{>Zü = [5, 3] and c = 0.6.

These set of values has ghZ

:[<? = 27 and R2 = 0.53, meaning

that, if compared to the former approach, it provides about

60% more voxels albeit keeping similar statistical precision.

As for data sets in Table 1, the mean value of the fractional

error c = skPL / kPL in a single data set varies between 0.16 and

0.34, giving an idea of how they link to the graphic outcomes

of fitting in Fig. 3.

In Fig. 4 we show examples of 5 kPL maps superimposed to

FLAIR images of a patient affected with gliosarcoma (data set

14) and of 5 kPB maps superimposed to T1-weighted images of

a patient affected with gliosarcoma (data set 16).

Figure 4. [Top] Examples of kPL maps superimposed to FLAIR images of a

patient (data set 14) affected with glioma (active tumor highlighted in red)

[Bottom] Examples of kPB maps superimposed to T1-weighted images of a

patient (data set 16). Sinc interpolation and the error criterion (skPL <

0.6?kPL) were used to generate kinetic maps in both cases (see text for detail).

All the kPL maps have been computed by using the error

criterion with ?SNR:;<

{>Z, SNR=>?

{>Zü = [5, 3] and skPL < 0.6?kPL.

The patient affected with gliosarcoma has previously received

radiation and temozolomide chemotherapy, along with

adjuvant agent pembrolizumab.

The interpretation of the correspondence between higher or

lower kPL values and presence of tumor is deferred to future

studies since it is beyond the scope of this communication,

that is mainly focused on methods to compute kinetic maps for

our brain data sets.

V. CONCLUSIONS

We conducted preliminary studies of hyperpolarized [1-

13C]

pyruvate injected in 10 patients affected with glioma,

demonstrating the feasibility for quantifying kinetic rates kPL

and kPB, which describe the conversion of pyruvate into lactate

and bicarbonate. We compared five kinetic models and two

approaches for discarding voxels with improper fitting and we

provided quantitative and spatially resolved maps of kinetic

rates. Such hyperpolarized MRI experiments enable an

unprecedented window into cerebral metabolism, which is not

provided by other molecular imaging techniques.

REFERENCES AND FOOTNOTES

1. Ardenkjaer-Larsen, J. H. On the present and future of

dissolution-DNP. Journal of Magnetic Resonance 264, 3–12

(2016).

2. Ardenkjaer-Larsen, J. H., Fridlund, B., Gram, A., Hansson, G.,

Hansson, L., Lerche, M. H., Servin, R., Thaning, M. & Golman,

K. Increase in signal-to-noise ratio of > 10,000 times in liquidstate NMR. Proceedings of the National Academy of Sciences

100, 10158–10163 (2003).

3. Mammoli, D., Vuichoud, B., Bornet, A., Milani, J., Dumez, J.-

N., Jannin, S. & Bodenhausen, G. Hyperpolarized para -

Ethanol. The Journal of Physical Chemistry B 119, 4048–4052

(2015).

極T代謝磁共振全球科研集錦

201

0278-0062 (c) 2019 IEEE. Personal use is permitted, but republication/redistribution requires IEEE permission. See http://www.ieee.org/publications_standards/publications/rights/index.html for more information.

This article has been accepted for publication in a future issue of this journal, but has not been fully edited. Content may change prior to final publication. Citation information: DOI 10.1109/TMI.2019.2926437, IEEE

Transactions on Medical Imaging

4. Bornet, A., Ji, X., Mammoli, D., Vuichoud, B., Milani, J.,

Bodenhausen, G. & Jannin, S. Long-Lived States of

Magnetically Equivalent Spins Populated by Dissolution-DNP

and Revealed by Enzymatic Reactions. Chemistry - A European

Journal 20, 17113–17118 (2014).

5. Bowen, S. & Hilty, C. Rapid sample injection for

hyperpolarized NMR spectroscopy. Physical Chemistry

Chemical Physics 12, 5766 (2010).

6. Day, S. E., Kettunen, M. I., Gallagher, F. A., Hu, D.-E., Lerche,

M., Wolber, J., Golman, K., Ardenkjaer-Larsen, J. H. &

Brindle, K. M. Detecting tumor response to treatment using

hyperpolarized 13C magnetic resonance imaging and

spectroscopy. Nature Medicine 13, 1382–1387 (2007).

7. Ardenkjaer-Larsen, J.-H., Boebinger, G. S., Comment, A.,

Duckett, S., Edison, A. S., Engelke, F., Griesinger, C., Griffin,

R. G., Hilty, C., Maeda, H., Parigi, G., Prisner, T., Ravera, E.,

van Bentum, J., Vega, S., Webb, A., Luchinat, C., Schwalbe, H.

& Frydman, L. Facing and Overcoming Sensitivity Challenges

in Biomolecular NMR Spectroscopy. Angewandte Chemie

International Edition 54, 9162–9185 (2015).

8. Buratto, R., Bornet, A., Milani, J., Mammoli, D., Vuichoud, B.,

Salvi, N., Singh, M., Laguerre, A., Passemard, S., GerberLemaire, S., Jannin, S. & Bodenhausen, G. Drug Screening

Boosted by Hyperpolarized Long-Lived States in NMR.

ChemMedChem 9, 2509–2515 (2014).

9. Lee, Y., Zeng, H., Ruedisser, S., Gossert, A. D. & Hilty, C.

Nuclear Magnetic Resonance of Hyperpolarized Fluorine for

Characterization of Protein–Ligand Interactions. Journal of the

American Chemical Society 134, 17448–17451 (2012).

10. Vasos, P. R., Comment, A., Sarkar, R., Ahuja, P., Jannin, S.,

Ansermet, J.-P., Konter, J. A., Hautle, P., van den Brandt, B. &

Bodenhausen, G. Long-lived states to sustain hyperpolarized

magnetization. Proceedings of the National Academy of

Sciences 106, 18469–18473 (2009).

11. Jannin, S., Bornet, A., Melzi, R. & Bodenhausen, G. High field

dynamic nuclear polarization at 6.7T: Carbon-13 polarization

above 70% within 20min. Chemical Physics Letters 549, 99–

102 (2012).

12. Meier, B., Dumez, J.-N., Stevanato, G., Hill-Cousins, J. T., Roy,

S. S., H?kansson, P., Mamone, S., Brown, R. C. D., Pileio, G. &

Levitt, M. H. Long-Lived Nuclear Spin States in Methyl Groups

and Quantum-Rotor-Induced Polarization. Journal of the

American Chemical Society 135, 18746–18749 (2013).

13. Pileio, G., Bowen, S., Laustsen, C., Tayler, M. C. D., HillCousins, J. T., Brown, L. J., Brown, R. C. D., ArdenkjaerLarsen, J. H. & Levitt, M. H. Recycling and Imaging of Nuclear

Singlet Hyperpolarization. Journal of the American Chemical

Society 135, 5084–5088 (2013).

14. Tayler, M. C. D., Marco-Rius, I., Kettunen, M. I., Brindle, K.

M., Levitt, M. H. & Pileio, G. Direct Enhancement of Nuclear

Singlet Order by Dynamic Nuclear Polarization. Journal of the

American Chemical Society 134, 7668–7671 (2012).

15. Gallagher, F. A., Kettunen, M. I., Day, S. E., Hu, D.-E.,

Ardenkj?r-Larsen, J. H., Zandt, R. in ‘t, Jensen, P. R., Karlsson,

M., Golman, K., Lerche, M. H. & Brindle, K. M. Magnetic

resonance imaging of pH in vivo using hyperpolarized 13Clabelled bicarbonate. Nature 453, 940–943 (2008).

16. Gallagher, F. A., Kettunen, M. I. & Brindle, K. M. Biomedical

applications of hyperpolarized 13C magnetic resonance

imaging. Progress in Nuclear Magnetic Resonance

Spectroscopy 55, 285–295 (2009).

17. Chen, A. P., Albers, M. J., Cunningham, C. H., Kohler, S. J.,

Yen, Y.-F., Hurd, R. E., Tropp, J., Bok, R., Pauly, J. M.,

Nelson, S. J., Kurhanewicz, J. & Vigneron, D. B.

Hyperpolarized C-13 spectroscopic imaging of the TRAMP

mouse at 3T—Initial experience. Magnetic Resonance in

Medicine 58, 1099–1106 (2007).

18. von Morze, C., Larson, P. E. Z., Hu, S., Keshari, K., Wilson, D.

M., Ardenkjaer-Larsen, J. H., Goga, A., Bok, R., Kurhanewicz,

J. & Vigneron, D. B. Imaging of blood flow using

hyperpolarized [13C]Urea in preclinical cancer models. Journal

of Magnetic Resonance Imaging 33, 692–697 (2011).

19. Comment, A. Dissolution DNP for in vivo preclinical studies.

Journal of Magnetic Resonance 264, 39–48 (2016).

20. Keshari, K. R., Sriram, R., Koelsch, B. L., Van Criekinge, M.,

Wilson, D. M., Kurhanewicz, J. & Wang, Z. J. Hyperpolarized

13C-Pyruvate Magnetic Resonance Reveals Rapid Lactate

Export in Metastatic Renal Cell Carcinomas. Cancer Research

73, 529–538 (2013).

21. Marco-Rius, I., von Morze, C., Sriram, R., Cao, P., Chang, G.-

Y., Milshteyn, E., Bok, R. A., Ohliger, M. A., Pearce, D.,

Kurhanewicz, J., Larson, P. E. Z., Vigneron, D. B. & Merritt,

M. Monitoring acute metabolic changes in the liver and kidneys

induced by fructose and glucose using hyperpolarized

[2-13C]dihydroxyacetone: Monitoring Metabolic Changes with

[2-13C]dihydroxyacetone. Magnetic Resonance in Medicine 77,

65–73 (2017).

22. Nelson, S. J., Kurhanewicz, J., Vigneron, D. B., Larson, P. E.

Z., Harzstark, A. L., Ferrone, M., van Criekinge, M., Chang, J.

W., Bok, R., Park, I., Reed, G., Carvajal, L., Small, E. J.,

Munster, P., Weinberg, V. K., Ardenkjaer-Larsen, J. H., Chen,

A. P., Hurd, R. E., Odegardstuen, L.-I., Robb, F. J., Tropp, J. &

Murray, J. A. Metabolic Imaging of Patients with Prostate

Cancer Using Hyperpolarized [1-13C]Pyruvate. Science

Translational Medicine 5, 198ra108-198ra108 (2013).

23. Cunningham, C. H., Lau, J. Y. C., Chen, A. P., Geraghty, B. J.,

Perks, W. J., Roifman, I., Wright, G. A. & Connelly, K. A.

Hyperpolarized 13C Metabolic MRI of the Human

HeartNovelty and Significance: Initial Experience. Circulation

Research 119, 1177–1182 (2016).

24. Aggarwal, R., Vigneron, D. B. & Kurhanewicz, J.

Hyperpolarized [1-13C]-Pyruvate Magnetic Resonance Imaging

Detects an Early Metabolic Response to Androgen Ablation

Therapy in Prostate Cancer. European Urology 72, 1028–1029

(2017).

25. Park, I., Larson, P. E. Z., Gordon, J. W., Carvajal, L., Chen, H.-

Y., Bok, R., Van Criekinge, M., Ferrone, M., Slater, J. B., Xu,

D., Kurhanewicz, J., Vigneron, D. B., Chang, S. & Nelson, S. J.

Development of methods and feasibility of using hyperpolarized

carbon-13 imaging data for evaluating brain metabolism in

patient studies: Hyperpolarized Carbon-13 Metabolic Imaging

of Patients With Brain Tumors. Magnetic Resonance in

Medicine 80, 864–873 (2018).

26. Chen, H.-Y., Larson, P. E. Z., Gordon, J. W., Bok, R. A.,

Ferrone, M., van Criekinge, M., Carvajal, L., Cao, P., Pauly, J.

M., Kerr, A. B., Park, I., Slater, J. B., Nelson, S. J., Munster, P.

N., Aggarwal, R., Kurhanewicz, J. & Vigneron, D. B.

Technique development of 3D dynamic CS-EPSI for

hyperpolarized 13C pyruvate MR molecular imaging of human

prostate cancer. Magnetic Resonance in Medicine (2018).

doi:10.1002/mrm.27179

27. Gordon, J. W., Hansen, R. B., Shin, P. J., Feng, Y., Vigneron,

D. B. & Larson, P. E. Z. 3D hyperpolarized C-13 EPI with

calibrationless parallel imaging. Journal of Magnetic Resonance

289, 92–99 (2018).

28. Miloushev, V. Z., Granlund, K. L., Boltyanskiy, R.,

Lyashchenko, S. K., DeAngelis, L. M., Mellinghoff, I. K.,

Brennan, C. W., Tabar, V., Yang, T. J., Holodny, A. I., Sosa, R.

E., Guo, Y. W., Chen, A. P., Tropp, J., Robb, F. & Keshari, K.

R. Metabolic Imaging of the Human Brain with Hyperpolarized

極T代謝磁共振全球科研集錦

202

0278-0062 (c) 2019 IEEE. Personal use is permitted, but republication/redistribution requires IEEE permission. See http://www.ieee.org/publications_standards/publications/rights/index.html for more information.

This article has been accepted for publication in a future issue of this journal, but has not been fully edited. Content may change prior to final publication. Citation information: DOI 10.1109/TMI.2019.2926437, IEEE

Transactions on Medical Imaging

13C Pyruvate Demonstrates 13C Lactate Production in Brain

Tumor Patients. Cancer Research 78, 3755–3760 (2018).

29. Albers, M. J., Bok, R., Chen, A. P., Cunningham, C. H.,

Zierhut, M. L., Zhang, V. Y., Kohler, S. J., Tropp, J., Hurd, R.

E., Yen, Y.-F., Nelson, S. J., Vigneron, D. B. & Kurhanewicz, J.

Hyperpolarized 13C Lactate, Pyruvate, and Alanine:

Noninvasive Biomarkers for Prostate Cancer Detection and

Grading. Cancer Research 68, 8607–8615 (2008).

30. Brindle, K. M., Bohndiek, S. E., Gallagher, F. A. & Kettunen,

M. I. Tumor imaging using hyperpolarized 13C magnetic

resonance spectroscopy. Magnetic Resonance in Medicine 66,

505–519 (2011).

31. Golman, K., Zandt, R. i., Lerche, M., Pehrson, R. & ArdenkjaerLarsen, J. H. Metabolic Imaging by Hyperpolarized 13C

Magnetic Resonance Imaging for In vivo Tumor Diagnosis.

Cancer Research 66, 10855–10860 (2006).

32. Lau, A. Z., Chen, A. P., Ghugre, N. R., Ramanan, V., Lam, W.

W., Connelly, K. A., Wright, G. A. & Cunningham, C. H. Rapid

multislice imaging of hyperpolarized 13C pyruvate and

bicarbonate in the heart. Magnetic Resonance in Medicine 64,

1323–1331 (2010).

33. Merritt, M. E., Harrison, C., Storey, C., Jeffrey, F. M., Sherry,

A. D. & Malloy, C. R. Hyperpolarized 13C allows a direct

measure of flux through a single enzyme-catalyzed step by

NMR. Proceedings of the National Academy of Sciences 104,

19773–19777 (2007).

34. Schroeder, M. A., Cochlin, L. E., Heather, L. C., Clarke, K.,

Radda, G. K. & Tyler, D. J. In vivo assessment of pyruvate

dehydrogenase flux in the heart using hyperpolarized carbon-13

magnetic resonance. Proceedings of the National Academy of

Sciences 105, 12051–12056 (2008).

35. Golman, K., Petersson, J. S., Magnusson, P., Johansson, E.,

?keson, P., Chai, C.-M., Hansson, G. & M?nsson, S. Cardiac

metabolism measured noninvasively by hyperpolarized13C

MRI. Magnetic Resonance in Medicine 59, 1005–1013 (2008).

36. Fuetterer, M., Busch, J., Peereboom, S. M., von Deuster, C.,

Wissmann, L., Lipiski, M., Fleischmann, T., Cesarovic, N.,

Stoeck, C. T. & Kozerke, S. Hyperpolarized 13C urea

myocardial first-pass perfusion imaging using velocity-selective

excitation. Journal of Cardiovascular Magnetic Resonance 19,

(2017).

37. Warburg, O. On the origin of cancer cells. Science 123, 309–

314 (1956).

38. McLean, M. A., Daniels, C., Grist, J., Schulte, R., Lanz, T.,

Chhabra, A., Earl, H. M., Basu, B., Wilkinson, I. B., Lomas, D.,

Caldas, C., Abraham, J. E., Graves, M., Gilbert, F., Brindle, K.

M. & Gallagher, F. A. Feasibility of metabolic imaging of

hyperpolarized 13C-pyruvate in human breast cancer. (2018).

doi:10.17863/CAM.21154

39. Zhu, Z. & et al. Development of Hyperpolarized 13C Dynamic

Breath-held MR Techniques and Initial Feasibility Studies for

Metabolic Imaging of Liver Metastases. Proceeding of Joint

Annual Meeting ISMRM-ESMRMB (2018).

40. Gallagher, F. A., Bohndiek, S. E., Kettunen, M. I., Lewis, D. Y.,

Soloviev, D. & Brindle, K. M. Hyperpolarized 13C MRI and

PET: In Vivo Tumor Biochemistry. Journal of Nuclear

Medicine 52, 1333–1336 (2011).

41. Keshari, K. R., Wilson, D. M., Van Criekinge, M., Sriram, R.,

Koelsch, B. L., Wang, Z. J., VanBrocklin, H. F., Peehl, D. M.,

O’Brien, T., Sampath, D., Carano, R. A. D. & Kurhanewicz, J.

Metabolic response of prostate cancer to nicotinamide

phophoribosyltransferase inhibition in a hyperpolarized

MR/PET compatible bioreactor: Metabolic Dynamics in

MR/PET Bioreactor. The Prostate 75, 1601–1609 (2015).

42. Logothetis, N. K., Pauls, J., Augath, M., Trinath, T. &

Oeltermann, A. Neurophysiological investigation of the basis of

the fMRI signal. Nature 412, 150–157 (2001).

43. Bélanger, M., Allaman, I. & Magistretti, P. J. Brain energy

metabolism: focus on astrocyte-neuron metabolic cooperation.

Cell Metab. 14, 724–738 (2011).

44. Zierhut, M. L., Yen, Y.-F., Chen, A. P., Bok, R., Albers, M. J.,

Zhang, V., Tropp, J., Park, I., Vigneron, D. B., Kurhanewicz, J.,

Hurd, R. E. & Nelson, S. J. Kinetic modeling of hyperpolarized

13C1-pyruvate metabolism in normal rats and TRAMP mice.

Journal of Magnetic Resonance 202, 85–92 (2010).

45. Bankson, J. A., Walker, C. M., Ramirez, M. S., Stefan, W.,

Fuentes, D., Merritt, M. E., Lee, J., Sandulache, V. C., Chen,

Y., Phan, L., Chou, P.-C., Rao, A., Yeung, S.-C. J., Lee, M.-H.,

Schellingerhout, D., Conrad, C. A., Malloy, C., Sherry, A. D.,

Lai, S. Y. & Hazle, J. D. Kinetic Modeling and Constrained

Reconstruction of Hyperpolarized [1-13C]-Pyruvate Offers

Improved Metabolic Imaging of Tumors. Cancer Research 75,

4708–4717 (2015).

46. Harrison, C., Yang, C., Jindal, A., DeBerardinis, R. J.,

Hooshyar, M. A., Merritt, M., Dean Sherry, A. & Malloy, C. R.

Comparison of kinetic models for analysis of pyruvate-to-lactate

exchange by hyperpolarized 13C NMR: comparison of models

of LDH exchange for 13C hyperpolarized puryvate. NMR in

Biomedicine 25, 1286–1294 (2012).

47. Hill, D. K., Orton, M. R., Mariotti, E., Boult, J. K. R., Panek,

R., Jafar, M., Parkes, H. G., Jamin, Y., Miniotis, M. F., AlSaffar, N. M. S., Beloueche-Babari, M., Robinson, S. P., Leach,

M. O., Chung, Y.-L. & Eykyn, T. R. Model Free Approach to

Kinetic Analysis of Real-Time Hyperpolarized 13C Magnetic

Resonance Spectroscopy Data. PLoS ONE 8, e71996 (2013).

48. Kazan, S. M., Reynolds, S., Kennerley, A., Wholey, E., Bluff, J.

E., Berwick, J., Cunningham, V. J., Paley, M. N. & Tozer, G.

M. Kinetic modeling of hyperpolarized 13C pyruvate

metabolism in tumors using a measured arterial input function:

Kinetic Modeling of 13C Pyruvate Metabolism. Magnetic

Resonance in Medicine 70, 943–953 (2013).

49. Mariotti, E., Orton, M. R., Eerbeek, O., Ashruf, J. F., Zuurbier,

C. J., Southworth, R. & Eykyn, T. R. Modeling non-linear

kinetics of hyperpolarized [1- 13C] pyruvate in the crystalloidperfused rat heart: Modelling non-linear kinetics of

hyperpolarized pyruvate. NMR in Biomedicine 29, 377–386

(2016).

50. Park, J. M., Josan, S., Jang, T., Merchant, M., Yen, Y.-F., Hurd,

R. E., Recht, L., Spielman, D. M. & Mayer, D. Metabolite

kinetics in C6 rat glioma model using magnetic resonance

spectroscopic imaging of hyperpolarized [1- 13C]pyruvate.

Magnetic Resonance in Medicine 68, 1886–1893 (2012).

51. Witney, T. H., Kettunen, M. I. & Brindle, K. M. Kinetic

Modeling of Hyperpolarized 13C Label Exchange between

Pyruvate and Lactate in Tumor Cells. Journal of Biological

Chemistry 286, 24572–24580 (2011).

52. Gordon, J. W., Chen, H.-Y., Autry, A., Park, I., Van Criekinge,

M., Mammoli, D., Milshteyn, E., Bok, R., Xu, D., Li, Y.,

Aggarwal, R., Chang, S., Slater, J. B., Ferrone, M., Nelson, S.,

Kurhanewicz, J., Larson, P. E. Z. & Vigneron, D. B. Translation

of Carbon-13 EPI for hyperpolarized MR molecular imaging of

prostate and brain cancer patients. Magnetic Resonance in

Medicine 81, 2702–2709 (2019).

53. Larson, P. E. Z., Chen, H.-Y., Gordon, J. W., Korn, N.,

Maidens, J., Arcak, M., Tang, S., Criekinge, M., Carvajal, L.,

Mammoli, D., Bok, R., Aggarwal, R., Ferrone, M., Slater, J. B.,

Nelson, S. J., Kurhanewicz, J. & Vigneron, D. B. Investigation

of analysis methods for hyperpolarized 13C-pyruvate metabolic

MRI in prostate cancer patients: Hyperpolarized Pyruvate

極T代謝磁共振全球科研集錦

203

0278-0062 (c) 2019 IEEE. Personal use is permitted, but republication/redistribution requires IEEE permission. See http://www.ieee.org/publications_standards/publications/rights/index.html for more information.

This article has been accepted for publication in a future issue of this journal, but has not been fully edited. Content may change prior to final publication. Citation information: DOI 10.1109/TMI.2019.2926437, IEEE

Transactions on Medical Imaging

Prostate Cancer Analysis Methods. NMR in Biomedicine 31,

e3997 (2018).

54. Ardenkjaer-Larsen, J. H., Leach, A. M., Clarke, N., Urbahn, J.,

Anderson, D. & Skloss, T. W. Dynamic nuclear polarization

polarizer for sterile use intent. NMR in Biomedicine 24, 927–

932 (2011).

55. Azema, M., Carvajal, L., Xu, D., Gordon, J. W., Park, I.,

Vigneron, D. B., Nelson, S. J., Stockmann, J. P., Keil, B. &

Wald, L. L. 31-Channel Brain Array for Hyperpolarized 13C

Imaging at 3T. ISMRM Abstract#1225 (2017).

56. Pruessmann, K. P., Weiger, M., B?rnert, P. & Boesiger, P.

Advances in sensitivity encoding with arbitrary k -space

trajectories: SENSE With Arbitrary k -Space Trajectories.

Magnetic Resonance in Medicine 46, 638–651 (2001).

57. Maidens, J., Gordon, J. W., Chen, H.-Y., Park, I., Van

Criekinge, M., Milshteyn, E., Bok, R., Aggarwal, R., Ferrone,

M., Slater, J. B., Kurhanewicz, J., Vigneron, D. B., Arcak, M. &

Larson, P. E. Z. Spatio-temporally constrained reconstruction

for hyperpolarized carbon-13 MRI using kinetic models. IEEE

Transactions on Medical Imaging 1–1 (2018).

doi:10.1109/TMI.2018.2844246

58. Gordon, J. W., Niles, D. J., Adamson, E. B., Johnson, K. M. &

Fain, S. B. Application of flow sensitive gradients for improved

measures of metabolism using hyperpolarized 13C MRI:

Improved Measures of Metabolism with 13C MRI. Magnetic

Resonance in Medicine 75, 1242–1248 (2016).

59. Maidens, J., Gordon, J. W., Arcak, M. & Larson, P. E. Z.

Optimizing Flip Angles for Metabolic Rate Estimation in

Hyperpolarized Carbon-13 MRI. IEEE Transactions on Medical

Imaging 35, 2403–2412 (2016).

60. Xing, Y., Reed, G. D., Pauly, J. M., Kerr, A. B. & Larson, P. E.

Z. Optimal variable flip angle schemes for dynamic acquisition

of exchanging hyperpolarized substrates. Journal of Magnetic

Resonance 234, 75–81 (2013).

61. Santarelli, M. F., Positano, V., Giovannetti, G., Frijia, F.,

Menichetti, L., Ardenkjaer-Larsen, J.-H., De Marchi, D.,

Lionetti, V., Aquaro, G., Lombardi, M. & Landini, L. How the

signal-to-noise ratio influences hyperpolarized 13C dynamic

MRS data fitting and parameter estimation: SNR ON 13C

DYNAMIC MRS. NMR in Biomedicine 25, 925–934 (2012).

62. Sun, C., Walker, C. M., Michel, K. A., Venkatesan, A. M., Lai,

S. Y. & Bankson, J. A. Influence of parameter accuracy on

pharmacokinetic analysis of hyperpolarized pyruvate:

Pharmacokinetic Analysis of Hyperpolarized Pyruvate.

Magnetic Resonance in Medicine 79, 3239–3248 (2018).

63. Granlund, K. L. & et al. Utilizing hyperpolarized MRI in

prostate cancer to assess metabolic dynamics and

histopathologic grade. Proceeding of Joint Annual Meeting

ISMRM-ESMRMB (2017).

64. Grist, J. T., McLean, M. A., Riemer, F., Schulte, R. F., Deen, S.

S., Zaccagna, F., Woitek, R., Daniels, C. J., Kaggie, J. D.,

Matys, T., Patterson, I., Slough, R., Gill, A. B., Chhabra, A.,

Eichenberger, R., Laurent, M.-C., Comment, A., Gillard, J. H.,

Coles, A. J., Tyler, D. J., Wilkinson, I., Basu, B., Lomas, D. J.,

Graves, M. J., Brindle, K. M. & Gallagher, F. A. Quantifying

normal human brain metabolism using hyperpolarized [1–

13C]pyruvate and magnetic resonance imaging. NeuroImage

189, 171–179 (2019).

極T代謝磁共振全球科研集錦

204

Lactate topography of the human brain using hyperpolarized 13C-MRI

研究背景

研究結(jié)果

研究對(duì)象

???????????代謝??????????? - ?????????????????????? ATP ??

??????????????????????研????????????????????????????

????????????????????????????????研????????????????????

???????代謝??????研????極?] 1-13C]pyruvate 代謝???????????????代謝??

??? ???????????????????

?I???II???? 29 ?? 69 ?????代謝?? [

13C]lactate ?] 13C]bicarbonate ??????????????

???????????????????????????????ǖ?A??????ǘ?B?????ǘ?C?

?????????

?III?? [

13C]lactate?A??] 13C]bicarbonate?B??????? ??z-score ???

14????ǖ???????? ????代謝??

I II III

極T代謝磁共振全球科研集錦

205

研究結(jié)論

應(yīng)用方向

[1-13C]pyruvate代謝??[

13C]lactate?]13C]bicarbonate???????????????????????????

??????????????代謝???????研?????代謝?????研??????????????

??????代謝?????????研??

?????代謝???????????????????????????

IEEE Trans Med Imaging. 2020 Feb;39(2):320-327.

doi: 10.1109/TMI.2019.2926437.

極T代謝磁共振全球科研集錦

206

Journal Pre-proof

Lactate topography of the human brain using hyperpolarized 13C-MRI

Casey Y. Lee, Hany Soliman, Benjamin J. Geraghty, Albert P. Chen, Kim A. Connelly,

Ruby Endre, William J. Perks, Chris Heyn, Sandra E. Black, Charles H. Cunningham

PII: S1053-8119(19)30793-1

DOI: https://doi.org/10.1016/j.neuroimage.2019.116202

Reference: YNIMG 116202

To appear in: NeuroImage

Received Date: 27 March 2019

Revised Date: 19 August 2019

Accepted Date: 16 September 2019

Please cite this article as: Lee, C.Y., Soliman, H., Geraghty, B.J., Chen, A.P., Connelly, K.A., Endre, R.,

Perks, W.J., Heyn, C., Black, S.E., Cunningham, C.H., Lactate topography of the human brain using

hyperpolarized 13C-MRI, NeuroImage (2019), doi: https://doi.org/10.1016/j.neuroimage.2019.116202.

This is a PDF file of an article that has undergone enhancements after acceptance, such as the addition

of a cover page and metadata, and formatting for readability, but it is not yet the definitive version of

record. This version will undergo additional copyediting, typesetting and review before it is published

in its final form, but we are providing this version to give early visibility of the article. Please note that,

during the production process, errors may be discovered which could affect the content, and all legal

disclaimers that apply to the journal pertain.

? 2019 Published by Elsevier Inc.

極T代謝磁共振全球科研集錦

207

極T代謝磁共振全球科研集錦

208

Lactate Topography of the Human

Brain using Hyperpolarized 13C-MRI

Casey Y. Lee1,2, Hany Soliman2, Benjamin J. Geraghty8,1, Albert P. Chen3,

Kim A. Connelly4, Ruby Endre8, William J. Perks5, Chris Heyn6, Sandra E.

Black7, and Charles H. Cunningham8,1

1Department of Medical Biophysics, University of Toronto, Toronto, Ontario, Canada

2Radiation Oncology, Sunnybrook Health Sciences Centre, Toronto, ON, Canada 3GE Healthcare, Toronto, Ontario, Canada 4 Keenan Research Centre, Li Ka Shing Knowledge Institute, St. Michael’s Hospital,

Toronto, ON, Canada 5Pharmacy, Sunnybrook Health Sciences Centre, Toronto, ON, Canada

6Radiology, Sunnybrook Health Sciences Centre, Toronto, ON Canada 7Department of Medicine (Neurology) and Hurvitz Brain Sciences Research Program,

Sunnybrook Health Sciences Centre, Toronto, ON Canada 8Physical Sciences, Sunnybrook Research Institute, Toronto, Ontario, Canada

Running Head: Lactate Topography of the Human Brain

Contact Information:

Charles H. Cunningham (Corresponding Author)

Department of Medical Biophysics, University of Toronto.

2075 Bayview Avenue, Toronto, Ontario M4N 3M5, Canada.

Tel: (416) 899-4406

Email: charles.cunningham@utoronto.ca

Wordcount (Main Article): 2138

極T代謝磁共振全球科研集錦

209

2

Abstract

Lactate is now recognized as an important intermediate in brain metabolism, but its role is

still under investigation. In this work we mapped the distribution of lactate and bicarbonate

produced from intravenously injected 13C-pyruvate over the whole brain using a new imaging

method, hyperpolarized 13C MRI (N=14, ages 23 to 77). Segmenting the 13C-lactate images

into brain atlas regions revealed a pattern of lactate that was preserved across individuals.

Higher lactate signal was observed in cortical grey matter compared to white matter and was

highest in the precuneus, cuneus and lingual gyrus. Bicarbonate signal, indicating flux of

[1-13C]pyruvate into the TCA cycle, also displayed consistent spatial distribution. One-way

ANOVA to test for significant differences in lactate among atlas regions gave F = 87.6 and p

< 10?6. This report of a “l(fā)actate topography” in the human brain and its consistent pattern

is evidence of region-specific lactate biology that is preserved across individuals.

Key Words

Lactate; Bicarbonate; Hyperpolarized 13C MRI; Metabolism; Structural Segmentation; Aerobic Glycolysis

Abbreviations

ANLS astrocyte-neuron lactate shuttle

OGI oxygen to glucose index

DE-EPI dual-echo echo-planar imaging

ANOVA one-way analysis of variance

W coefficient of Kendall’s concordance

LDH lactate dehydrogenase

rCBF regional cerebral blood flow

極T代謝磁共振全球科研集錦

210

3

1 Introduction

Lactate, once considered a waste product, is now recognized as an important intermediate

in brain metabolism (1, 2, 3, 4). The astrocyte-neuron lactate shuttle (ANLS) model, first

proposed by Pellerin and Magistretti (1), is that the ATP needed to clear glutamate from

synapses is primarily derived from lactate produced by aerobic glycolysis in astrocytes. There

is evidence of multiple roles for lactate: a source of energy in glutamatergic neurons, a

signalling molecule modulating neuronal excitability and synaptic plasticity, and a key player

in maintaining homeostasis (4). Furthermore, experiments in rats have shown that lactate

transport is required for long-term memory formation (5), and that cerebral aerobic glycolysis

and lactate concentration are reduced during sleep (6).

In this work, we mapped the distribution of lactate and bicarbonate produced from

intravenously injected 13C-pyruvate over the whole brain using a new imaging method, hyperpolarized 13C MRI. The 13C-lactate images were segmented into brain atlas regions, which

revealed a consistent spatial distribution of lactate across individuals. This is the first report

of such a “l(fā)actate topography” in the human brain. The consistent pattern is evidence of

region-specific lactate biology that is preserved across individuals.

While there continues to be debate surrounding the degree to which neurons are fueled by

astrocytic lactate (7, 8), mounting evidence points to a critical role of aerobic glycolysis and

lactate in brain energy metabolism. Combined PET measurements of cerebral metabolic rate

of oxygen and glucose using 15O-oxygen and 19F-fluorodeoxyglucose, respectively, can provide

an estimate of aerobic glycolysis through a measure of the oxygen to glucose index (OGI),

which is 6.0 if glucose is fully oxidized. These PET experiments have shown clear evidence of

non-oxidative consumption of glucose (OGI < 6.0), with net brain glucose uptake exceeding

the rate required to match the regional oxygen consumption. OGI is lowest in children and

young adults and increases with aging (9, 10), and recently, it was shown that the OGI

has a distinct topography within the brain, corresponding to regions with persistent gene

expression associated with childhood development (neoteny) (11, 12).

極T代謝磁共振全球科研集錦

211

4

Hyperpolarized 13C-MRI is a new imaging method that enables time-resolved volumetric

imaging of metabolite production within tissue (13) and has recently been translated to

humans (14, 15, 16). Using this method, we mapped the distribution of 13C-lactate and 13Cbicarbonate produced from intravenously injected 13C-pyruvate in the brains of volunteer

subjects (N=14).

2 Materials and Methods

Written informed consent was obtained from all subjects (N = 14) prior to study participation

under a protocol approved by the Sunnybrook Research Ethics Board and by Health Canada

as a Clinical Trial Application. A 20-gauge intravenous catheter was inserted into the forearm

of each subject before they were positioned supine and head-first in a GE MR750 3.0T

MRI scanner (GE Healthcare, Waukesha, WI). A custom 13C head coil was used to acquire

3D dual-echo echo-planar imaging (DE-EPI) data of [1-13C]lactate, [1-13C]bicarbonate, and

[1-13C]pyruvate (axial, FOV 24 x 24 x 36 cm3, 1.5 cm-isotropic resolution, 5 s temporal

resolution, total of 60 s acquisition). At each timepoint, lactate, bicarbonate, and pyruvate

images were acquired with net tip angles of 80?, 80?, 11?, respectively, which was spread

across 24 excitations. After the metabolic images, the 13C head coil was carefully replaced

with a standard 8-channel 1H neurovascular array (Invivo Inc.) and standard anatomical

images were acquired.

Each subject was injected with a 0.1 mmol/kg dose of [1-13C]pyruvate prepared within

a sterile fluid path. Each dose contained 1.47 grams of [1-13C]pyruvic acid (Sigma Aldrich,

St. Louis, MO) and AH111501 [Tris(8-carboxy-2,2,6,6 (tetra(methoxyethyl) benzo-[1,2-

4,5]bis-(1,3)dithiole-4-yl)methyl sodium salt] (Syncom, Groningen, The Netherlands) in a

49:1 weight by weight ratio, respectively. This mixture was then hyperpolarized in a polarizer (General Electric SPINLab system, equipped with a quality control module) for 3

hours to achieve maximum polarization. Just prior to 13C image acquisition, the sample

極T代謝磁共振全球科研集錦

212

5

was dissolved within the sterile fluid pathway by 38 mL of heated and pressurized sterile

water, transferred to the receiver vessel and mixed with 17.5 mL of a neutralizing solution

(600 mmol/L NaOH, 333 mmol/L Tris base, and 333 mg/L disodium EDTA) plus 19 mL of

sterile water. An aliquot of the final product was used for quality control assessment and

the remainder (45 mL) was transferred into a Medrad (Medrad, Indianola PA) syringe for

injection. The sample was injected at 4 mL/s followed by a 25 mL normal saline flush at 5

mL/s using a Spectris Solaris power injector. The 13C image acquisition was initiated at the

end of the saline flush.

Volumetric images of [1-13C]lactate, 13C-bicarbonate, and [1-13C]pyruvate were acquired

using a 3 Tesla MRI scanner and a 13C head coil. Metabolite images covering a 3D volume

(128 x 16 x 24 voxels / 1.5 cm isotropic resolution) were acquired at 5 s intervals over

the 60 s acquisition window. The center frequency of the spectrally-selective echo-planar

pulse sequence (17, 18) was toggled between the resonance frequencies of [1-13C]lactate, 13Cbicarbonate and [1-13C]pyruvate, resulting in separate volumetric images for each of these

metabolites, every 5 s. Following the metabolite images, a standard suite of anatomical brain

images was acquired with a conventional head coil.

To understand how the observed spatial distribution of metabolite signals related to brain

structure, the LONI pipeline processing environment (19) was used to segment the metabolite

images from each subject into the 56 regions contained in the LPBA40 atlas (20). The 13C image analysis workflow was modified from the LONI Brain Parser workflow. First, the 13C images were resampled to match the coordinates and matrix size of the anatomical T1-weighted

images (axial fast spoiled gradient echo images, FOV 25.6 x 25.6 cm2, 1 mm-isotropic resolution, TR/TE 7.6/2.9 ms, flip angle 11?). The T1-weighted images from each subject were

registered to the averaged human brain images used to produce the LPBA40 atlas by generating and applying a B-spline deformation field. The Brain Parser package was then used

on these forward deformation field-warped T1-weighted images to label the 56 regions. The

labelled 3D images were then warped by the inverse of the deformation field to transform the

極T代謝磁共振全球科研集錦

213

6

labels back to the coordinates of the original T1-weighted images. The mri segstats program

included in the Freesurfer package (http://surfer.nmr.mgh.harvard.edu) was then used to

compute mean signals from each atlas region from the [1-13C]lactate and 13C-bicarbonate

images from each subject. These values were normalized by converting to a z -score (the

number of standard deviations from the mean of all regions within a single subject). This

normalization accounts for any global scaling of the data for an individual subject, such

the signal-enhancement level of the injected 13C-pyruvate and any difference in substrate

delivery to the brain.

3 Results

Representative metabolite images from two subjects are shown in Fig. 1. Lactate signal,

which results from the conversion of [1-13C]pyruvate to [1-13C]lactate, was observed in all

subjects with a consistent topography. Higher lactate signal was observed in cortical grey

matter compared to white matter and was highest in the precuneus, cuneus and lingual

gyrus. Bicarbonate signal, which indicates the flux of [1-13C]pyruvate through the pyruvatedehydrogenase complex on the mitochondrial membrane, resulting in 13CO2 which is rapidly

converted to 13C-bicarbonate, had a similarly consistent spatial distribution, and was also

highest in many of the same regions as lactate, such as the occipital lobe.

The segmented regional z -score analysis revealed a highly consistent but spatially variable

pattern of lactate signal across subjects, as seen in Fig. 2(a). One-way analysis of variance

(ANOVA) was run to test for significant differences in lactate z-score among atlas regions,

giving F = 87.6 and p < 10?6. The same analysis reported the significant differences in

bicarbonate z -score among atlas regions, giving F = 79.6 and p < 10?6. Consistent regional

distribution patterns were observed across subjects for all metabolite z-scores, with Kendall’s

concordance coefficients (W ) of 0.83 for lactate, 0.82 for bicarbonate and 0.85 for pyruvate.

The lactate-to-pyruvate ratio for each region displayed large variance between subjects

極T代謝磁共振全球科研集錦

214

7

(see supplemental Fig. 6(a)). However, the concordance of the regional lactate-to-pyruvate

ratio (W = 0.75) was only slightly lower that for regional lactate (W = 0.83), showing a

fairly consistent pattern in the lactate-to-pyruvate ratio across subjects. Plotting the lactate

z-scores vs. the pyruvate z-scores for all subjects showed that the lactate and pyruvate

signals were correlated until a lactate z-score around 1.0, and were not correlated for the

higher lactate regions (see supplemental Fig. 6(d)).

To further investigate whether cerebral venous drainage is dominating the regional variation in 13C-lactate signal, a skilled neuroradiologist drew ROIs on the right jugular vein

in a subset of the subjects (N=5), and the timecourse of the 13C-lactate and 13C-pyruvate

in this ROI was plotted (see Supplementary Fig. 7). The observed lactate-to-pyruvate ratio

(computed as an area-under-the-curve ratio) was significantly lower in the jugular ROIs as

compared with all of the brain atlas regions, which is evidence against the 13C-lactate signal

observed in the brain being dominated by venous signal. A limitation of this analysis is

that the jugular veins are immediately adjacent to the carotid arteries, so contamination of

the jugular signal with 13C-pyruvate signal from the artery could have biased the lactateto-pyruvate ratio downwards. However, the lactate signal in the jugular ROIs was still far

lower than the lactate signal in the surrounding brain parenchyma, which has much lower

blood volume, with all of the jugular lactate z -scores below 0 (-2.65 to -0.37). As for 13Cbicarbonate, the literature on hyperpolarized 13C-pyruvate to date contains no reports of

13C-bicarbonate signal in the blood. This widely observed finding is consistent with 13Cbicarbonate being trapped in the intracellular space on the timescale of the hyperpolarized

13C experiment. Thus it is unlikely that the 13C-bicarbonate signal observed in the brain is

in the vascular pool.

The potential for bias in the regional metabolite signals due to the 1.5 cm (isotropic)

voxel size in the 13C images was investigated. Region volumes for the 14 subjects are plotted

in Fig. 5 (supplementary). Kendall’s concordance (W ) coefficient for the region volumes confirmed strong agreement between subjects (W = 0.98). The concordance (W ) between the

極T代謝磁共振全球科研集錦

215

8

mean lactate z -scores and mean regional volumes was 0.59, confirming a weak concordance.

Regional volumes ranged between 1.40 - 104.03 cm3 (median 11.96 cm3). Of 56 segmented

regions, 48 regional volumes were larger than the volume of a single isotropic 1.5 cm voxel.

4 Discussion

The high 13C-lactate signal level observed here is consistent with the 13C-lactate observed

in recent data from other groups (16, 21, 22). While the steady-state lactate concentration

in the brain measured by proton spectroscopy is relatively low (0.3 mM to 1 mM) (23,

24) and increases only slightly with activation (25, 26), the hyperpolarization makes 13Clactate readily observable at sub-millimolar concentration. The consistency of the lactate

topography across subjects suggests that region-specific lactate biology is similar between

individuals.

A number of previous studies have reported the regional distribution of factors that

are potentially related to the observed 13C-lactate pattern in the human brain including

the regional distribution of aerobic glycolysis (11, 9, 27), the enzyme lactate deydrogenase

(LDH) (28, 29), cerebral blood flow (30), oxidative damage to mitochondrial DNA (31) and

FDG uptake (32). There is some agreement between these distributions and the high lactate

regions seen here. For example, the topography of elevated aerobic glycolysis (reduced OGI)

(11, 9) showed elevated aerobic glycolysis in the precuneus, and regional cerebral blood

flow (rCBF) maps show the high z-scores in both the lingual gyrus and precuneus (30).

However, two of the regions reported to have elevated aerobic glycolysis corresponded to

the regions with the lowest lactate z -scores (caudate and gyrus rectus), and the superior

frontal gyrus shows high FDG uptake (32) but had a mean lactate z-score near zero, so

more investigation is required to understand how these factors may be related. It is worth

noting that reduced OGI indicates regions where there is a net non-oxidative use of glucose,

whereas hyperpolarized MRI shows all lactate created from the injected substrate, including

極T代謝磁共振全球科研集錦

216

9

lactate that ultimately ends up being used in oxidative phosphorylation.

The appearance of [1-13C]lactate signal is dependent on a different set of glycolysis-related

factors in comparison to the uptake of radiolabeled glucose and oxygen. These major factors are the expression levels of monocarboxylate transporters to transport pyruvate and

lactate across lipid bilayers, the availability of the enzyme LDH, and the local concentration

of NADH, which is oxidized to NAD+ when pyruvate is reduced to lactate in the cytosol.

Depletion of NADH has been shown to be the dominant factor in the 13C-lactate signal reduction observed after treating cancer cells (33), so regional variation in NADH concentration

is a possible explanation for the observed lactate pattern.

5 Conclusion

Hyperpolarized 13C MRI revealed a consistent lactate topography across subjects of varying

age. This consistent pattern is evidence of region-specific lactate biology that is preserved

across individuals.

6 Figures and Tables

極T代謝磁共振全球科研集錦

217

10

4924

2462 1266

2529

1937

3873

968

1935

875

1750

1750

3500

2750

5500

1500

3000

Lactate Bicarbonate

Subject #10 (29 y.o. ) Subject #9 (69 y.o.)

A

B

C

Figure 1: Representative images of 13C-lactate (left) and 13C-bicarbonate (right) from

healthy female volunteers aged 29 (upper) and 69 (lower). The metabolite signals are displayed as colour overlays on the corresponding T1-weighted anatomical images in grayscale,

and were computed by summing the 12 timepoints over the 60 s acquisition window. The

arrows indicate (A) the left precuneus, (B) the left cuneus and (C) the reference sample used

for pre-scan calibration.

極T代謝磁共振全球科研集錦

218

11

A

B

Figure 2: The topography of normalized lactate and bicarbonate signals across subjects

(N=14). (A) Lactate and (B) bicarbonate z -scores plotted vs. the LPBA40 atlas region

labels, with each colour showing a different subject.

極T代謝磁共振全球科研集錦

219

12

A

B

Figure 3: Consensus maps of the (A) lactate and (B) bicarbonate signals calculated from

the mean z -score for each LPBA40 atlas region.

極T代謝磁共振全球科研集錦

220

13

A

B

Figure 4: Heatmaps representing the lactate and bicarbonate z-scores with the regions in

ascending order for lactate from one of the subjects (23,M). LPBA40 numbers and names

of the regions with the 8 highest lactate z -scores are: (49) left precuneus, (67) left cuneus,

(50) right precuneus, (68) right cuneus, (90) right lingual gyrus, (122) right cingulate gyrus,

(121) left cingulate gyrus, (89) left lingual gyrus.

極T代謝磁共振全球科研集錦

221

Quantifying normal human brain metabolism

using hyperpolarized [1-13C]pyruvate and magnetic resonance imaging

研究背景

研究結(jié)果

研究對(duì)象

???????代謝??????????????????????????????????????????

???????????????????????? A?????????????????????????

???????????代謝???????????????????1H-MRS??????????????

????????????

?極?] 1-13C]pyruvate ?????????????????????研???研?????極?] 1-13C]pyruvate ?

???????????????????????????????代謝??????????????????

? (I):????代謝?]?1-13C]pyruvate?171ppm??代謝?]?

13C]???183ppm??[13C]?????161ppm????

???????????集????????4s????????????????????????????

? (II): ??????????????????????????????????????????????

? (III): ???????????????????????代謝?kPL?kPB???????????????????

??????????????????????

??[1-13C]pyruvate????????????13C?????????????????????????????

??????????????????代謝???????代謝????????????????????

?????????????????????????????????????????????????

?????????????????????????????????????代謝???kPL???0.012±

0.006s-1?kPB???0.002± 0.002s-1????13C?????????? ????????????

4??????ǖ???????代謝

極T代謝磁共振全球科研集錦

222

研究結(jié)論

應(yīng)用方向

??????極?] 1-13C]pyruvate MRI ???????????????代謝????????????????

代謝???????????????????????????????????????????????

?????代謝??? ?

? ?????????????????研?

? ??????????????研?

極T代謝磁共振全球科研集錦

223

Quantifying normal human brain metabolism using hyperpolarized [1–13C]

pyruvate and magnetic resonance imaging

James T. Grist a, Mary A. McLean b, Frank Riemer a, Rolf F. Schulte c, Surrin S. Deen a,b,

Fulvio Zaccagna a, Ramona Woitek a,b, Charlie J. Daniels a, Joshua D. Kaggie a, Tomasz Matyz a,

Ilse Patterson d

, Rhys Slough d, Andrew B. Gill a, Anita Chhabra g, Rose Eichenberger h,

Marie-Christine Laurent a, Arnaud Comment b,i, Jonathan H. Gillard a, Alasdair J. Colesj,

Damian J. Tyler k

, Ian Wilkinson e, Bristi Basu f, David J. Lomas a, Martin J. Graves d,

Kevin M. Brindle b, Ferdia A. Gallagher a,*

a Department of Radiology, University of Cambridge, Cambridge, UK

b Cancer Research UK Cambridge Institute, University of Cambridge, Cambridge, UK

c General Electric Healthcare, Munich, Germany

d Radiology, Cambridge University Hospitals, Cambridge, UK

e Department of Medicine, University of Cambridge and Cambridge Clinical Trials Unit, Cambridge University Hospitals NHS Foundation Trust, Cambridge, UK f Department of Oncology, University of Cambridge, Cambridge, UK

g Pharmacy, Cambridge University Hospitals, Cambridge, UK

h University of Cambridge, MRC Epidemiology Unit, Cambridge, UK

i GE Healthcare, Chalfont St Giles, UK

j Department of Clinical Neurosciences, University of Cambridge, Cambridge, UK

k Department of Physiology, Anatomy, and Genetics, University of Oxford, Oxford, UK

ARTICLE INFO

Keywords:

Metabolism

Hyperpolarized

MRI

Carbon-13

Brain

Pyruvate

ABSTRACT

Hyperpolarized 13C Magnetic Resonance Imaging (13C-MRI) provides a highly sensitive tool to probe tissue

metabolism in vivo and has recently been translated into clinical studies. We report the cerebral metabolism of

intravenously injected hyperpolarized [1–13C]pyruvate in the brain of healthy human volunteers for the first time.

Dynamic acquisition of 13C images demonstrated 13C-labeling of both lactate and bicarbonate, catalyzed by

cytosolic lactate dehydrogenase and mitochondrial pyruvate dehydrogenase respectively. This demonstrates that

both enzymes can be probed in vivo in the presence of an intact blood-brain barrier: the measured apparent exchange rate constant (kPL) for exchange of the hyperpolarized 13C label between [1–13C]pyruvate and the

endogenous lactate pool was 0.012 ! 0.006 s"1 and the apparent rate constant (kPB) for the irreversible flux of

[1–13C]pyruvate to [13C]bicarbonate was 0.002 ! 0.002 s"1

. Imaging also revealed that [1–13C]pyruvate, [1–13C]

lactate and [13C]bicarbonate were significantly higher in gray matter compared to white matter. Imaging normal

brain metabolism with hyperpolarized [1–13C]pyruvate and subsequent quantification, have important implications for interpreting pathological cerebral metabolism in future studies.

1. Introduction

Cerebral metabolism is important for normal brain function and becomes deranged in a number of pathological processes, such as inflammation, infection, ischemia, traumatic brain injury and in tumors (Jalloh

et al., 2015; Mathur et al., 2014; Matz et al., 2006). 18F-fluorodeoxyglucose (FDG) uptake, detected using positron emission tomography

(PET), is one approach to imaging this cerebral metabolism in patients.

Despite the sensitivity of PET, the signal acquired from 18F-FDG represents flux in only part of the glycolytic pathway measuring a combination

of delivery to the tissue, uptake by glucose transporters and subsequent

phosphorylation in the reaction catalyzed by the glycolytic enzyme,

hexokinase. As the technique cannot detect downstream products of

glucose metabolism, such as lactate and CO2, it provides no direct

* Corresponding author. Department of Radiology, School of Clinical Medicine, Box 218, University of Cambridge, Cambridge, CB2 0QQ, UK.

E-mail address: fag1000@cam.ac.uk (F.A. Gallagher).

Contents lists available at ScienceDirect

NeuroImage

journal homepage: www.elsevier.com/locate/neuroimage

https://doi.org/10.1016/j.neuroimage.2019.01.027

Received 12 September 2018; Received in revised form 8 January 2019; Accepted 10 January 2019

Available online 11 January 2019

1053-8119/? 2019 Published by Elsevier Inc.

NeuroImage 189 (2019) 171–179

極T代謝磁共振全球科研集錦

224

information on glycolytic fluxes and mitochondrial oxidative metabolism. The work presented here uses a new imaging method to investigate cerebral metabolism of pyruvate, a breakdown product of glucose.

Pyruvate is transported across both the intact blood-brain barrier and

the plasma membrane by the monocarboxylate transporters (MCTs).

Pyruvate can either be metabolized to lactate, catalyzed by cytosolic

lactate dehydrogenase (LDH), or is metabolized by oxidative decarboxylation to acetyl-CoA, catalyzed by mitochondrial pyruvate dehydrogenase (PDH). Proton (1

H) magnetic resonance spectroscopy (MRS) of the

healthy brain has demonstrated steady state cerebral lactate concentrations in the region of 0.6–1 mM, although it may be up to 2.2 mM in

neonates, where glucose metabolism is altered compared to the adult

brain (Prichard et al., 1991; Tomiyasu et al., 2016). The metabolic shift

from mitochondrial oxidative metabolism to glycolysis and lactate formation occurs in a number of pathological processes, such as ischemia,

inflammation, and in tumors (Jalloh et al., 2015; Mathur et al., 2014;

Matz et al., 2006). However, imaging of the spatial and temporal distribution of lactate by 1

H-MRS is inhibited by a low signal-to-noise ratio

(SNR) at clinical field strengths. Therefore, alternative non-invasive

methods to image lactate would be valuable to monitor glycolysis in vivo.

Hyperpolarized carbon-13 Magnetic Resonance Imaging (13C-MRI)

has emerged as a promising technique for studying tissue metabolism in

humans (Zaccagna et al., 2018). The method increases the SNR of liquid

state carbon-13 MRS (13C-MRS) by more than four orders of magnitude

(Ardenkjaer-Larsen et al., 2011; Hurd et al., 2012). This substantial increase in SNR has been used to non-invasively image the spatial distribution of intravenously injected 13C-labelled molecules in vivo such as

[1–13C]pyruvate. Importantly, dynamic 13C-MRI acquisition allows the

injected hyperpolarized [1–13C]pyruvate to be differentiated from its

metabolic products such as [1–13C]lactate, [1–13C]alanine and 13C-labelled carbon dioxide/bicarbonate as they form in real-time. There

are a number of challenges for imaging hyperpolarized carbon-13

metabolism, particularly the short half-life of the hyperpolarized

signal, which is typically 25–30 s for [1–13C]lactate and [1–13C]pyruvate

in vivo but as low as 10 s for [13C]bicarbonate (Gallagher et al., 2008).

This limits the range of metabolites that can be probed using this technique. Furthermore, the hyperpolarized signal is irreversibly depleted as

images are acquired and therefore efficient imaging strategies are

required to maximize the data that can be obtained. There are a number

of sequences that have been used in human studies and here we have

utilized IDEAL (Iterative Decomposition with Echo Asymmetry and Least

squares estimation) spiral chemical shift imaging (CSI), as it allows for

spatial averaging to increase the SNR for short lived metabolites (Gordon

et al., 2016; Wiesinger et al., 2012).

In order to quantify the dynamics of pyruvate metabolism, a number

of quantitative approaches for describing the exchange of hyperpolarized

carbon-13 label between pyruvate and lactate have been proposed,

including both model-based and model-free methods (Gomez Dami ! !an

et al., 2014; Khegai et al., 2014; Schulte et al., 2013). Modelling the

process as a two-site exchange system, which gives the apparent exchange rate constant (kPL) for label flux between pyruvate and lactate, is

the most accurate approach. However, time-to-peak (TTP) for the lactate

signal intensity and the ratio of the integrals of the lactate and pyruvate

signals (area under the curve, AUC) are simple model-free approaches

that can also be used to estimate label flux (Daniels et al., 2016).

Quantifying these metrics in normal tissue is important so that changes in

diseased tissue can be understood and monitored over time. Frequency

domain kinetic modelling has been shown to be robust in low SNR environments, such as hyperpolarized 13C-MRI data. Furthermore, the

frequency domain benefits from incorporating the arterial input function

(AIF) into the data, which would otherwise be challenging to accurately

estimate from the low spatial resolution hyperpolarized images (Khegai

et al., 2014).

Previous studies in rodent and non-human primate brains have

demonstrated cerebral lactate labelling following injection of hyperpolarized [1–13C]pyruvate, with a kPL of 0.0026 s!1 reported in the

macaque brain, albeit using a higher dose than currently used for humans

(~0.38 mmol/kg compared to ~0.11 mmol/kg) (Park et al., 2014).

Although formation of hyperpolarized [13C]bicarbonate has also been

observed in some rodent studies using a high pyruvate dose, it has not

been reported in non-human primates. This suggests that cerebral pyruvate metabolism may be dose dependent, that there may be interspecies variation in pyruvate metabolism, or that metabolism may be

affected by anaesthesia (Josan et al., 2013).

Hyperpolarized [1–13C]pyruvate has recently been applied to patient

studies with the first report in prostate cancer (Nelson et al., 2013). More

recent studies have demonstrated the technique in normal human heart

and as a treatment response marker in prostate cancer (Aggarwal et al.,

2017; Cunningham et al., 2016). Lactate labelling has also been

demonstrated in patients with brain tumors following therapy, where

there is also preliminary evidence for the formation of cerebral bicarbonate (Park et al., 2018; Miloushev et al., 2018). However, as these

tumors are highly invasive and have undergone therapy, the metabolism

of normal brain has not yet been established. Here we have used 13C-MRI

to image the conversion of hyperpolarized [1–13C]pyruvate into both

[1–13C]lactate and [13C]bicarbonate in the normal human brain and

have quantified pyruvate metabolism in gray and white matter.

2. Method and materials

2.1. Subject recruitment and screening

Local ethical approval was obtained for this prospective study (NRES

Committee East of England, Cambridge South, REC number 15/EE/

0255). Between September 2017 and March 2018, four volunteers (mean

age 27 " 2 years, one male, three female) were consented and screened

prior to imaging; this included assessment of blood pressure, oxygen

saturation, heart rate, electrocardiogram (ECG), and blood analysis (urea

& electrolytes, full blood count, serum lactate, serum glucose, and lactate

dehydrogenase (LDH)). Only volunteers with normal screening tests

were included in the study. Blood sampling was undertaken prior to

imaging and 30 min after. Oxygen saturation and heart rate were monitored during pyruvate injection and throughout the examination. The

volunteers were observed for up to 30 min after the end of the

examination.

2.2. 13C pyruvate preparation

Pharmacy kits/fluid paths for insertion into the clinical hyperpolarizer (SPINlab, 5T, Research Circle Technology, Niskayuna, NY)

were filled under sterile conditions. 1.47g [1–13C]pyruvic acid (Sigma

Aldrich, St Louis, Missouri, USA) containing 15 mM of an electron

paramagnetic agent (EPA, Syncom, Groningen, Netherlands) was sealed

in a vial; 38 mL sterile water was used for dissolution; 19 mL sterile water

with 17.5 mL NaOH/Tris/EDTA (2.4%, 4.03%, and 0.033% w/v

respectively, Royal Free Hospital, London) was used as a buffer for

neutralisation. Pharmacy kits were stored in a freezer at !20 #C for at

least two weeks prior to use (Zaccagna et al., 2018). The vial containing

the frozen pyruvate/EPA mix was defrosted in the helium pressurised

airlock in the hyperpolarizer for one hour. The sample was irradiated at

139 GHz at ~0.8 K for approximately three hours. Following rapid

dissolution, the pyruvic acid was neutralised with the buffer and Quality

Control (QC) checks were performed by an integrated QC module which

measured: pyruvate and EPA concentration, pH, temperature, sample

polarisation and volume of dissolute. The release criteria for injection

were: pyruvate concentration 220–280 mM; radical concentration

<3 μM; pH 6.7–8.1; and temperature 25–37 #C. After release, the sample

was passed through a hatch into the adjacent MRI scanner room and

0.4 mL/kg of the final ~250 mM hyperpolarized pyruvate solution was

injected at 5 mL/s using a syringe driver (Medrad, Warrendale, Pennsylvania, USA) followed by a saline flush of 25mL at 5 mL/s.

J.T. Grist et al. NeuroImage 189 (2019) 171–179

172

極T代謝磁共振全球科研集錦

225

2.3. Phantom imaging protocol

Imaging was undertaken using a 3T MR system (MR750, GE

Healthcare, Waukesha WI), using a dual-tuned 1

H/13C quadrature head

coil (Rapid Biomedical, Rimpar Germany).

Transmit Gain (TG) and center frequency (f0) were determined using

a Bloch-Siegert method (Schulte et al., 2011). To assess the 13C transmit

and receive B1 sensitivity of the coil, a uniform 16 cm diameter sphere

filled with pure polydimethylsiloxane was placed within a coil-loading

ring filled with saline (GE, GE Healthcare, Waukesha WI) inside of the

coil; this was used to obtain signal from natural abundance carbon

(Supplementary Fig. 1A). Transmit B1 was determined by acquiring 13C

IDEAL spiral CSI images with 9 nominal flip angles between 60! and 180!

and fitting a sine function to each voxel in the resulting images to

determine the ratio of nominal to actual flip angle (8-step cycle interleaving one slice-selective spectrum and seven spirals, Field of View

(FOV) 400 mm, slice thickness 40 mm, acquired resolution 26 " 26 mm2

,

reconstructed resolution 8.3 mm2, flip angles 70–180! in 10! increments,

repetition time (TR) 1 s, echo shift 1.1 ms) (Wiesinger et al., 2012).

2.4. Clinical imaging protocol

The clinical imaging was performed with the same MR system and

coil set up as for the phantom imaging.

The clinical 1

H imaging protocol comprised: T1-weighted volumetric

imaging (3D; inversion prepared gradient echo; inversion time ? 450 ms;

FOV ? 240 mm; TR ? 8.6 ms; echo time (TE) ? 3.3 ms; flip angle

(FA) ? 12!; spatial resolution ? 0.9 " 0.9 " 1 mm3

); B0 field map

(FOV ? 240 mm, TE ? 7, 14 ms, TR ? 100 ms, FA ? 20!, spatial resolution ? 1 " 1x5 mm3

). 13C transmit gain (TG) and center frequency (f0) were set using a13C

enriched urea phantom (8 M, Sigma-Aldrich, UK) attached to the ear

defenders worn by the subject. 13C imaging was undertaken using a dynamic IDEAL spiral acquisition (pulse bandwidth ? 2500 Hz, TR ? 0.5 s;

time resolution ? 4 s; FA ? 15!; FOV ? 240 mm; acquired spatial resolution ? 12 " 12 mm2

; reconstructed resolution ? 5 " 5 mm2; slice thickness ? 30 mm, acquired voxel volume ? 4.32 cm3

, total imaging time

60 s). Images and spectra were reconstructed with 15 Hz line broadening.

Data acquisition began 10 s after the end of injection. Summed images at

the acquired resolution without zero-filling are shown in Supplementary

Fig. 2C to demonstrate the true resolution. 13C image reconstruction, post-processing, and analysis were performed in Matlab (The Mathworks, Natick, MA). Data and postprocessing code are available upon request to the authors.

2.5. Quantitative post processing

Imaging data were reconstructed by explicitly calculating the IDEAL

Fourier matrix, prior to inversion. B0 maps were summed over each 13C

imaging slab. An example B0 map is shown in Supplementary Fig. 2B. B0

correction was applied during inversion, using an additional frequency

demodulation component (Moriguchi et al., 2003).

The slice spatial offset between metabolites was defined by equation

(1):

Δz ? Δf

γGss

(1)

Where Δz is the spatial shift (m), Δf the frequency difference between

metabolites (Hz), γ the gyromagnetic ratio of 13C (MHzT$1

), and Gss the

strength of the slice-select gradient (mT). This offset was used to determine the separate range of thin axial imaging slices contributing signal to

each metabolite individually.

Imaging and spectroscopic data were summed in the complex and

magnitude domains respectively, and ratio maps of lactate-to-pyruvate,

bicarbonate-to-pyruvate, and bicarbonate-to-lactate were calculated.

Total pyruvate, lactate, and bicarbonate maps were generated by

normalizing all the voxels to the peak pyruvate signal in the brain. The

rate constant, kPL, was calculated using a two-site exchange model using a

frequency-domain approach and linear least-squares fitting, with any

back conversion (kLP) and spin lattice relaxation effects combined as an

effective relaxation term, T1 eff (Khegai et al., 2014). The rate constant,

kPB, was also evaluated using a two-site model in the frequency domain,

representing the metabolism of [1–13C]pyruvate to [13C]carbon dioxide

catalyzed by pyruvate dehydrogenase (PDH), followed by exchange with

[

13C]bicarbonate, catalyzed by the enzyme carbonic anhydrase (Gomez

!

Dami!an et al., 2014).

dMBetT

dt ? $ρBMBetT t kPBMBetT (2)

Where MBetT is the time dependent bicarbonate signal, ρB is the

effective relaxivity of the bicarbonate signal (the inverse of T1 eff) and kPB

is the metabolic conversion rate of pyruvate to bicarbonate.

2.6. Image analysis

Segmented white, gray and whole brain matter masks were produced

from the 3D T1 weighted acquisition using statistical parametric mapping

(SPM12, Wellcome Trust Centre for Neuroimaging, UCL, London). A twostage approach was used to account for the chemical shift displacement

between different metabolites. Firstly, gray and white matter probability

maps were calculated by summing over different ranges of thin axial

imaging slices to match the thickness of the 13C imaging slices, offset for

each metabolite by its chemical shift displacement (Fig. 1A–B). Secondly,

binary maps were produced from these images which contained >60%

gray matter, white matter or brain for all three metabolites (Chard et al.,

2002).

2.7. Region of interest analysis

A number of regions of interest (ROIs) within the brain were evaluated to determine if there was spatial heterogeneity in tissue metabolism:

basal ganglia, deep white matter, corpus callosum, cortical gray matter,

and the brain stem (Fig. 2) to assess for spatial metabolic heterogeneity.

Several of these regions contained a combination of both gray and white

matter. Analysis was performed on the averaged voxels from all the

volunteers.

Tissue probability maps for each 13C slice were first calculated by

summing over a range of thin axial slices for each metabolite determined

by its chemical shift displacement from the transmitted frequency, as

illustrated for white matter in the superior slice of one volunteer. A:

lactate; B: pyruvate; C: bicarbonate. D: The final binary masks were

calculated where the average probability for gray matter or white matter

was >60% for all three metabolites; see text for details. E: 13C-pyruvate, 13C-lactate and 13C-bicarbonate distribution derived from these segmentation maps showing signal in white (unfilled) and gray (filled)

matter; signals are normalized to the peak pyruvate signal in the whole

brain, *p < 0.05.

A, B, and C (left to right): Example ROIs containing deep white

matter, basal ganglia, cortical gray matter, corpus callosum, and the

brain stem.

Whole brain, gray matter and white matter analyses were performed

with segmented tissue masks for total pyruvate, total lactate, total bicarbonate, kPL, kPB, lactate-to-pyruvate, bicarbonate-to-lactate, and

bicarbonate-to-pyruvate ratios by averaging all voxels acquired from all

volunteers using the segmented masks. Inter-slice gray and white matter

analyses were performed by averaging voxels from all volunteers on a

slice-by-slice basis. Further comparisons between metabolic parameters

were made between tissue regions of interest, as well as analysis with and

without B0 correction.

J.T. Grist et al. NeuroImage 189 (2019) 171–179

173

極T代謝磁共振全球科研集錦

226

2.8. Statistical analysis

Statistical analysis was undertaken by comparing paired values between gray and white matter on an inter-slice basis using the Wilcoxon

sign rank test in the Matlab Statistics and Machine Learning Toolbox. All

statistical results were corrected for multiple comparisons using a Bonferroni correction. Statistical significance was defined as p < 0.05.

3. Results

3.1. Coil radiofrequency homogeneity

The assessment of radiofrequency (RF) excitation (B1) uniformity

using the polydimethylsiloxane phantom demonstrated a highly homogeneous B1 field. The mean ratio of the nominal to the actual flip angle

within the central slice was 84 ! 3% (mean ! S.D.). The imaging and

slice profile results are shown in Supplementary Figs. 2B and 2C.

3.2. Hyperpolarized imaging

The time taken for dissolution and QC was 35 s. The time between the

release of the [1–13C]pyruvate filled syringe and the start of the intravenous injection was 11 ! 2 s (mean ! S.D.). The levels of polarisation

achieved in all four subjects, as measured in the liquid state by the QC

module, was 25 ! 3% (mean ! S.D.).

Summed spectra from the entire time course demonstrated [1–13C]

pyruvate signal (171 ppm) in the three axial slices acquired, which

extended from the brain vertex to the cerebellum (Fig. 3). These spectra

also demonstrated both [1–13C]llactate (183 ppm) and [13C]bicarbonate

(161 ppm) in all four volunteers and a small quantity of pyruvate hydrate

(177 ppm) was observed at early time points. Fig. 4A is an example

spectral time course with a time resolution of 4 s, demonstrating the

arrival/formation of the three metabolites over time in a single volunteer. Fig. 4B demonstrates the mean signal from all four volunteers,

normalized to the peak [1–13C]pyruvate signal in each case. On average,

signal from [1–13C]pyruvate, [1–13C]lactate and [13C]bicarbonate were

observed 4, 8, and 16 s after the start of imaging respectively, which

commenced 10 s after the start of the hyperpolarized [1–13C]pyruvate

injection.

A-C: The spatial location of the 3 cm 13C slices used in this study are

shown in green on a sagittal T1 weighted image through the brain: three

slices were imaged containing the cerebellum (inferior slice, A), basal

Fig. 1. The distribution of hyperpolarized signal from the three metabolites within gray and white matter.

Fig. 2. Region of interest analysis and a field map of the healthy brain.

J.T. Grist et al. NeuroImage 189 (2019) 171–179

174

極T代謝磁共振全球科研集錦

227

ganglia (central slice, B), and corona radiata (superior slice, C). D-F: show

the summed 13C magnitude spectra from the total time course acquisition, demonstrating signal from [1–13C]pyruvate (171 ppm), [1–13C]

lactate (183 ppm) and [13C]bicarbonate (161 ppm) in all slices.

A: Dynamic spectra acquired every 4 s from the central slice of a

volunteer, following injection of hyperpolarized [1–13C]pyruvate.

[1–13C]Pyruvate (171 ppm) inflow is seen with subsequent exchange into

[1–13C]lactate (183 ppm) at approximately 8 s after pyruvate arrival and

formation of [13C]bicarbonate (161 ppm) beginning at approximately

12 s. B: Average signal intensities (! S.D.) for all three metabolites from

all four volunteers demonstrating the temporal dynamics; signal has been

normalized to the peak [1–13C]pyruvate signal in each case. Lactate and

Fig. 3. 13C spectra acquired through the healthy brain following injection of hyperpolarized pyruvate.

Fig. 4. Dynamic 13C spectra from the healthy brain showing the time course of [1–13C]pyruvate, [1–13C]lactate and [13C]bicarbonate.

Fig. 5. IDEAL spiral 13C imaging demonstrating metabolite distribution in the healthy human brain.

J.T. Grist et al. NeuroImage 189 (2019) 171–179

175

極T代謝磁共振全球科研集錦

228

bicarbonate have been shown with a five-fold increase in signal intensity

for ease of viewing.

IDEAL spiral 13C-MRI demonstrated the spatial distribution of the

hyperpolarized metabolites throughout the brain (Fig. 5). Pyruvate signal

was observed in the cerebrum and cerebellum and in both gray and white

matter. The pyruvate and lactate signals were particularly high in the

cerebral venous sinuses (e.g. the superior sagittal sinus and at the

confluence of the sinuses), demonstrating that LDH activity and lactate

transport were sufficiently rapid to allow tissue washout during the

timescale of the experiment (see Supplementary Fig. 2A).

A. Example summed images from the brain of a healthy volunteer

(number 1) demonstrating [1–13C]pyruvate, [1–13C]lactate and [13C]

bicarbonate signal from three axial slices: superior, central and inferior.

The T1-weighted images have also been shown, as have the quantitative

maps of the exchange of pyruvate to lactate (kPL in s!1

). B: Similar imaging shown as in (A) from the central slice of the three other volunteers.

3.3. Tissue segmentation

Segmentation analysis of the whole brain is shown in Fig. 1E.

Significantly higher signal from all three metabolites was observed in

gray matter compared to white matter: [1–13C]pyruvate, 0.47 " 0.24 vs.

0.25 " 0.12; [1–13C]lactate, 0.09 " 0.04 vs. 0.06 " 0.03; [13C]bicarbonate, 0.03 " 0.01 vs. 0.02 " 0.01 respectively; p < 0.05). Since the difference in metabolites between gray and white matter was largest for

pyruvate, the ratios of lactate-to-pyruvate and bicarbonate-to-pyruvate

tended to be higher in white matter than gray, but this only reached

significance for lactate. There was no significant difference in kPL, kPB, or

bicarbonate-to-lactate ratio between the two tissue types (Fig. 6,

Table 1).

The mean kPL derived from the whole brain for all four subjects was

0.012 " 0.006 s!1

. Similar results were obtained for both segmented gray

and white matter (Table 1). The mean value for kPB derived from the

whole brain for all four subjects was 0.02 " 0.002 s!1

. Results from each

volunteer are shown in Supplementary Table 1. The effective mean pyruvate and lactate relaxation times (Tl eff) for the whole brain was

26 " 10 s.

A: Mean lactate-to-pyruvate ratios derived from the segmented imaging data for the three brain slices showing the signal from all voxels in

both white (unfilled) and gray (filled) matter averaged across all volunteers (mean " S.D.). B: Mean bicarbonate-to-pyruvate ratios. C:

Apparent exchange rate constants modelled from the time course data. D:

Mean bicarbonate-to-lactate ratios.

3.4. Region of interest analysis

In comparison, region of interest analysis revealed significant differences in kPL between deep white matter and regions containing basal

ganglia or the brainstem (0.008 " 0.002 vs. 0.024 " 0.05; 0.008 " 0.002

vs. 0.020 " 0.004 s!1

, respectively; p < 0.05 in both cases). This result

suggests that there are regional variations in kPL across the brain.

The 13C imaging data was used to derive quantitative parameters

from the whole brain, as well as segmented white and gray matter. Mean

(" S.D.) values for kPL, kPB, lactate-to-pyruvate ratio, bicarbonate-topyruvate ratio and bicarbonate-to-lactate ratio are shown.

3.5. Serum blood results

Serum analysis revealed an increase in lactate concentration between

baseline and 30 min after pyruvate injection: t24 " 8% (mean " S.D.;

n ? 3; range 0.1–0.7 mM). However, there was no change in serum

glucose or LDH. Volunteers experienced no change in baseline vital signs

and no significant side effects were experienced.

4. Discussion

Glucose, lactate and pyruvate all play a role as cerebral energy

sources. Astrocytic end-feet have high concentrations of glucose transporters and cover a large proportion of the capillary walls to facilitate

rapid glucose transport into the brain (Magistretti et al., 1999). Following

release by neurons, the neurotransmitter glutamate may undergo

sodium-dependent transport from the synaptic cleft into astrocytes,

where it stimulates glycolysis and lactate formation (Magistretti et al.,

1999). When this lactate is transported into the extracellular space by

MCTs, it may be taken up by neurons and converted into pyruvate, which

can then be used as an energy source. This hypothesis is supported by the

differential distribution of LDH isoforms between the two cell types:

LDH5 (comprising four LDHA subunits) has been shown to be present in

astrocytes but not neurons and is found in tissues that are more glycolytic, favoring the formation of lactate; in contrast, neurons express LDH1

exclusively (comprising four LDHB subunits) which is present in tissues

that have a predominately oxidative metabolism and favor the production of pyruvate (Bittar et al., 1996; Laughton et al., 2000). In this way,

there is a close metabolic coupling between astrocytes and neurons

involving an interplay between glucose, glutamate, pyruvate and lactate,

with astrocytes being more glycolytic and neurons being predominately

oxidative and consuming lactate (Bittar et al., 1996).

Although the roles of glucose and lactate in the brain are well

described, cerebral pyruvate transport and metabolism in the healthy

human brain is less well understood, as endogenous pyruvate concentrations are much lower and pyruvate is largely intracellular. The MCT

family transports pyruvate in addition to lactate; for example, astrocytes

express MCT1 and MCT4, and neurons express MCT2 (Pellerin et al.,

2007). MCT2 has a particularly high affinity for pyruvate with a Km of

0.1 mM, followed by MCT1 with a Km of 1.0 mM. Therefore, both cell

types, but particularly neurons, will rapidly transport pyruvate at the

peak tissue pyruvate concentrations achieved in the experiments

described here i.e. 0.1–1 mM (P!erez-Escuredo et al., 2016). The kinetics

of pyruvate metabolism observed in this study are a function of pyruvate

delivery to the brain, MCT expression, LDH activity and tissue lactate

concentration, all of which may vary between regions of the brain.

This study has quantified the metabolism of hyperpolarized pyruvate

in the healthy human brain for the first time. We have shown that [1–13C]

pyruvate is rapidly transported across the blood-brain barrier to form

[1–13C]lactate within the lifetime of the hyperpolarized signal. The peak

[1–13C]pyruvate and [1–13C]lactate signals were measured at 12 and 16 s

respectively, following the start of imaging. The presence of [1–13C]

lactate in the cerebral venous sinuses shows that there is also rapid

washout of the labelled lactate. The presence of [13C]bicarbonate

throughout the brain demonstrates that PDH activity is sufficient in the

normal human brain to enable mitochondrial function to be probed in

addition to cytosolic LDH activity; the peak signal from [13C]bicarbonate

was measured at 26 s following the end of injection. These results

demonstrate the possibility of applying this technology not only to diseases where lactate is elevated, but also as a biomarker of early mitochondrial damage, which is a feature of inflammation.

The signals acquired from [1–13C]pyruvate, [1–13C]lactate and [13C]

bicarbonate were higher in gray matter compared to white matter.

Perfusion differences between gray and white matter may partially account for the higher gray matter signal: gray matter perfusion has been

shown to be 1.4–4.0 times higher than in white matter (Li et al., 2014).

Given the relatively low temporal resolution of the metabolic imaging

used here, temporal differences in the pyruvate and lactate timecourses

could not be detected between gray and white matter. The lactate to

pyruvate ratio showed the only significant difference between tissue

types, which may be driven by the higher perfusion of gray matter.

Furthermore, although not significant in this N ? 4 population, there was

an increase in the bicarbonate-to-lactate ratio between the cortical gray

matter and deep white matter regions of interest, potentially showing

differences in metabolism. However, partial volume effects may also play

J.T. Grist et al. NeuroImage 189 (2019) 171–179

176

極T代謝磁共振全球科研集錦

229

a role as the thickness of the gray matter may be as small as 3 mm in

places and the results here represent a small sample size (Fischl and Dale,

2000). In comparison, region of interest analysis demonstrated a significant difference in kPL between areas of deep white matter and the basal

ganglia and brain stem, regions known to have high glycolytic activity

(Berti et al., 2014). This may be explained by regional variations in cerebral LDH expression or cellular density or cell type (Laughton et al.,

2000). Histochemical and in situ hybridisation methods have shown high

levels of LDH expression in the hippocampus, pons, thalamus and

neocortex of several species. Further larger studies, assessing repeatability and reproducibility of these findings and spatial heterogeneity

across the brain, will be important in understanding these preliminary

findings, as well as the development of methods to automatically

segment areas of high exchange by exploiting the multidimensional nature of dynamic hyperpolarized data and fully integrating spatial, temporal and spectral information (Daniels and Gallagher, 2017).

The mean whole brain kPL measured here was 0.012 ! 0.006 s"1

,

which is greater than a value of 0.0026 s"1 measured in the anaesthetised

macaque brain utilising a pyruvate dose which was approximately four

times greater per unit body weight (Park et al., 2014). A previous study

assessing the anesthetised porcine brain showed lactate formation only

after the transient opening of the blood-brain barrier, when a kPL of

0.012 ! 0.007 s"1 was measured which is similar to the values reported

here (Miller et al., 2018).. Higher kPL values have been measured in tumors, reflecting elevated LDH and lactate concentrations in these tissues.

For example, a value of 0.025 s"1 was reported in human prostate cancer

before treatment and 0.007 s"1 following androgen ablation therapy,

which is similar to the value reported here for normal tissue (Aggarwal

et al., 2017). However, the value of kPL measured here in normal brain is

lower than that found in a previous human brain tumour study, where a

mean whole brain kPL of 0.12 s"1 was detected; this included tumor as

well as normal-appearing brain and therefore kPL may be dominated by

the elevated tumor metabolism (Miloushev et al., 2018).

The mean kPB derived from the whole brain was 0.002 ! 0.002 s"1

,

which represents flux through the reaction catalyzed by PDH and subsequent exchange of the 13C label between carbon dioxide and bicarbonate. The latter reaction, which is catalyzed by carbonic anhydrase, is

rapid and assumed to be at equilibrium (Gallagher et al., 2015). Although

we were able to demonstrate higher bicarbonate signal in gray matter

compared to white matter, there were no significant tissue differences

found in kPB or in the ratios of bicarbonate-to-pyruvate or

lactate-to-pyruvate. The SNR of bicarbonate was a limiting factor in this

analysis acquisition and methods to increase this are important for future

studies: this could be achieved by incorporating spectral-spatial pulses to

selectively increase the excitation flip angle for bicarbonate (Larson et al.,

2008; Schulte et al., 2013). An increase in SNR will also enable acquisition of higher spatial resolution images of each metabolite. Future

larger studies are required to assess variation between individuals which

may be affected by factors such as age, gender and weight.

An important element of this study was the uniform RF excitation

profile, which has allowed a comparative quantitative analysis to be

undertaken across the brain. A uniform RF excitation profile is a

Fig. 6. Quantitative analysis of metabolism in white and gray matter.

Table 1

Quantitative metabolic parameters derived from regions within the brain without B0 correction. Segmented regions were automatically derived and regions of interest

(ROIs) were manually drawn.

kPL (s"1) kPB (s"1) Lactate: pyruvate ratio Bicarbonate: pyruvate ratio Bicarbonate: lactate ratio

Segmented whole brain mask 0.012 ! 0.006 0.002 ! 0.002 0.23 ! 0.07 0.07 ! 0.04 0.32 ! 0.15

Segmented white matter 0.012 ! 0.007 0.002 ! 0.002 0.25 ! 0.08 0.08 ! 0.05 0.32 ! 0.21

Segmented gray matter 0.011 ! 0.005 0.002 ! 0.002 0.22 ! 0.06 0.07 ! 0.03 0.32 ! 0.18

Cortical gray matter ROI 0.012 ! 0.001 0.003 ! 0.002 0.23 ! 0.02 0.08 ! 0.02 0.33 ! 0.1

Basal ganglia ROI 0.024 ! 0.005 0.002 ! 0.001 0.18 ! 0.03 0.04 ! 0.02 0.20 ! 0.1

Corpus callosum ROI 0.013 ! 0.004 0.002 ! 0.001 0.21 ! 0.03 0.07 ! 0.03 0.30 ! 0.2

Deep white matter ROI 0.008 ! 0.002 0.002 ! 0.001 0.22 ! 0.07 0.05 ! 0.02 0.20 ! 0.2

Brainstem ROI 0.020 ! 0.004 0.003 ! 0.002 0.22 ! 0.04 0.04 ! 0.02 0.21 ! 0.08

J.T. Grist et al. NeuroImage 189 (2019) 171–179

177

極T代謝磁共振全球科研集錦

230

consideration for selective excitation sequences, as the uncertainty in the

delivered flip angles may significantly affect the results derived from

kinetic modelling (Sun et al., 2018). The relative lack of contrast reported

here between kinetic parameters in normal-appearing gray and white

matter was unexpected, but could assist the identification of significantly

altered metabolism above or below background in pathological

conditions.

In conclusion, this study has demonstrated that 13C-MRI can be used

to acquire quantitative non-invasive measurements of hyperpolarized

[1–13C]pyruvate metabolism in the normal human brain and could be

used to measure regional variations in metabolism across the brain. This

work provides evidence that the methodology may have a role in

assessing disease processes where lactate is elevated and where mitochondrial function may be altered.

Acknowledgements

This study was funded by the Wellcome Trust, Cancer Research UK

(CRUK, C19212/A16628, C19212/A911376, C197/A16465), the CRUK

Cambridge Centre, National Institute of Health Research-Cambridge

Biomedical Research Centre, Medical Research Council, CRUK/Engineering and Physical Sciences Research Council Imaging Centre in

Cambridge and Manchester, Addenbrooke's Charitable Trust, the Cambridge Experimental Cancer Medicine Centre, the Evelyn Trust, the

Multiple Sclerosis Society, the National Institute for Health Research

[Cambridge Biomedical Research Centre at the Cambridge University

Hospitals NHS Foundation Trust and the Mark Foundation Institute for

Integrative Cancer Medicine at the University of Cambridge]. The views

expressed are those of the authors and not necessarily those of the NHS,

the NIHR or the Department of Health and Social Care.

Thanks to Sarah Hilborne, Jackie Mason, Vicky Fernandez, Hannah

Loveday, Ashley Grimmer, Emma Ward, Brian White, Amy Fray, Ronnie

Hernandez, Matthew Locke, Claire Trumper, Dario Prudencio, and Bruno

do Carmo.

Appendix A. Supplementary data

Supplementary data to this article can be found online at https://doi.

org/10.1016/j.neuroimage.2019.01.027.

References

Aggarwal, R., Vigneron, D.B., Kurhanewicz, J., 2017. Hyperpolarized 1-[13C]-pyruvate

Magnetic Resonance Imaging detects an early metabolic response to androgen

ablation therapy in prostate cancer. Eur. Urol. 72, 1028–1029.

Ardenkjaer-Larsen, J.H., Leach, A.M., Clarke, N., Urbahn, J., Anderson, D., Skloss, T.W.,

2011. Dynamic nuclear polarization polarizer for sterile use intent. NMR Biomed. 24,

927–932.

Berti, V., Mosconi, L., Pupi, A., 2014. Brain: normal variations and benign findings in FDG

PET/CT imaging. Pet. Clin. 14, 129–140.

Bittar, P.G., Charnay, Y., Pellerin, L., Bouras, C., Magistretti, P.J., 1996. Selective

distribution of lactate dehydrogenase isoenzymes in neurons. J. Cerebr. Blood Flow

Metabol. 16, 1079–1089.

Chard, D.T., Griffin, C.M., McLean, M.A., Kapeller, P., Kapoor, R., Thompson, A.J.,

Miller, D.H., 2002. Brain metabolite changes in cortical grey and normal-appearing

white matter in clinically early relapsing-remitting multiple sclerosis. Brain 125,

2342–2352.

Cunningham, C.H., Lau, J.Y.C., Chen, A.P., Geraghty, B.J., Perks, W.J., Roifman, I.,

Wright, G.A., Connelly, K.A., 2016. Hyperpolarized 13C metabolic MRI of the human

heart: initial experience. Circ. Res. 119, 1177–1182.

Daniels, C.J., Gallagher, F.A., 2017. Unsupervised segmentation of 5D hyperpolarized

carbon-13 MRI data using a fuzzy Markov random field model. IEEE Trans. Med.

Imag. 37, 840–850.

Daniels, C.J., McLean, M.A., Schulte, R.F., Robb, F.J., Gill, A.B., Mcglashan, N.,

Graves, M.J., Schwaiger, M., Lomas, D.J., Brindle, K.M., Gallagher, F.A., 2016.

A comparison of quantitative methods for clinical imaging with hyperpolarized. NMR

Biomed. 29, 387–399.

Fischl, B., Dale, A.M., 2000. Measuring the thickness of the human cerebral cortex from

magnetic resonance images. Proc. Natl. Acad. Sci. Unit. States Am. 97, 11050–11055.

Gallagher, F.A., Kettunen, M.I., Day, S.E., Hu, D.-E., Ardenkjaer-Larsen, J.H., Zandt, R. in

’T., Jensen, P.R., Karlsson, M., Golman, K., Lerche, M.H., Brindle, K.M., 2008.

Magnetic resonance imaging of pH in vivo using hyperpolarized 13C-labelled

bicarbonate. Nature 453, 940–943.

Gallagher, F.A., Sladen, H., Kettunen, M.I., Serrao, E.M., Rodrigues, T.B., Wright, A.,

Gill, A.B., McGuire, S., Booth, T.C., Boren, J., McIntyre, A., Miller, J.L., Lee, S.H.,

Honess, D., Day, S.E., Hu, D.E., Howat, W.J., Harris, A.L., Brindle, K.M., 2015.

Carbonic anhydrase activity monitored in vivo by hyperpolarized 13C-magnetic

resonance spectroscopy demonstrates its importance for pH regulation in tumors.

Cancer Res. 75, 4109–4118.

Gomez Dami ! !an, P.A., Sperl, J.I., Janich, M.A., Khegai, O., Wiesinger, F., Glaser, S.J.,

Haase, A., Schwaiger, M., Schulte, R.F., Menzel, M.I., 2014. Multisite kinetic

modeling of 13C metabolic MR using [1-13C]pyruvate. Radiol. Res. Pract. 871619,

2014.

Gordon, J.W., Vigneron, D.B., Larson, P.E.Z., 2016. Development of a symmetric echo

planar imaging framework for clinical translation of rapid dynamic hyperpolarized

13C imaging. Magn. Reson. Med. 77, 826–832.

Hurd, R.E., Yen, Y.-F., Chen, A., Ardenkjaer-Larsen, J.H., 2012. Hyperpolarized 13C

metabolic imaging using dissolution dynamic nuclear polarization. J. Magn. Reson.

Imag. 36, 1314–1328.

Jalloh, I., Carpenter, K.L.H., Helmy, A., Carpenter, T.A., Menon, D.K., Hutchinson, P.J.,

2015. Glucose metabolism following human traumatic brain injury: methods of

assessment and pathophysiological findings. Metab. Brain Dis. 30, 615–632.

Josan, S., Hurd, R., Billingsley, K., Senadheera, L., Park, J.M., Yen, Y.F., Pfefferbaum, A.,

Spielman, D., Mayer, D., 2013. Effects of isoflurane anesthesia on hyperpolarized 13C

metabolic measurements in rat brain. Magn. Reson. Med. 70, 1117–1124.

Khegai, O., Schulte, R.F., Janich, M.A., Menzel, M.I., Farrell, E., Otto, A.M., ArdenkjaerLarsen, J.H., Glaser, S.J., Haase, A., Schwaiger, M., Wiesinger, F., 2014. Apparent rate

constant mapping using hyperpolarized [1-13C]pyruvate. NMR Biomed. 27,

1256–1265.

Larson 1, P.E., Kerr, A.B., Chen, A.P., Lustig, M.S., Zierhut, M.L., Hu, S.,

Cunningham, C.H., Pauly, J.M., Kurhanewicz, J., Vigneron, D.B., 2008 Sep.

Multiband excitation pulses for hyperpolarized 13C dynamic chemical-shift imaging.

J. Magn. Res. 194 (1), 121–127.

Laughton, J.D., Charnay, Y., Belloir, B., Pellerin, L., Magistretti, P.J., Bouras, C., 2000.

Differential messenger RNA distribution of lactate dehydrogenase LDH-1 and LDH-5

isoforms in the rat brain. Neuroscience 96, 619–625.

Li, X., Sarkar, S.N., Purdy, D.E., Briggs, R.W., 2014. Quantifying cerebellum grey matter

and white matter perfusion using pulsed arterial spin labeling. BioMed Res. Int.

108691, 2014.

Magistretti, P.J., Pellerin, L., Rothman, D.L., Shulman, R.G., 1999. Energy on demand.

Science 283, 496–497.

Mathur, D., Rodas, G.L., Casanova, B., Marti, M.B., 2014. Perturbed glucose metabolism:

insights into multiple sclerosis pathogenesis. Front. Neurol. 5, 1–20.

Matz, K., Keresztes, K., Tatschl, C., Nowotny, M., Dachenhausen, A., Brainin, M.,

Tuomilehto, J., 2006. Disorders of glucose metabolism in acute stroke patients: an

underrecognized problem. Diabetes Care 29, 792–797.

Miller, J.J., Grist, J.T., Serres, S., Larkin, J.R., Lau, A.Z., Ray, K., Fisher, K.R., Hansen, E.,

Tougaard, R.S., Nielsen, P.M., Lindhardt, J., Laustsen, C., Gallagher, F.A., Tyler, D.J.,

Sibson, N., 2018. 13C pyruvate transport across the blood-brain barrier in preclinical

hyperpolarised MRI. Sci. Rep. 8, 15082.

Miloushev, V., Granlund, K.L., Boltyanskiy, R., Lyashchenko, S.K., DeAngelis, L.M.,

Mellinghoff, I.K., Brennan, C.W., Tabar, V., Yang, T.J., Holodny, A.I., Sosa, R.E.,

Guo, Y.W., Chen, A.P., Tro, J., Keshari, K., 2018. Metabolic imaging of the human

brain with hyperpolarized 13C pyruvate demonstrates 13C lactate production in brain

tumor patients. Cancer Res. 78, 3755–3760.

Moriguchi, H., Lewin, J.S., Duerk, J.L., 2003. Dixon techniques in spiral trajectories with

off-resonance correction: a new approach for fat signal suppression without spatialspectral RF pulses. Magn. Reson. Med. 50, 915–924.

Nelson, S.J., Kurhanewicz, J., Vigneron, D.B., Larson, P.E.Z., Harzstark, A.L., Ferrone, M.,

van Criekinge, M., Chang, J.W., Bok, R., Park, I., Reed, G., Carvajal, L., Small, E.J.,

Munster, P., Weinberg, V.K., Ardenkjaer-Larsen, J.H., Chen, A.P., Hurd, R.E.,

Odegards, L., Murray, J.A., 2013. Metabolic imaging of patients with prostate cancer

using hyperpolarized [1-13C]pyruvate. Sci. Transl. Med. 5, 198ra108.

Park, I., Larson, P.E.Z., Tropp, J.L., Carvajal, L., Reed, G., Bok, R., Robb, F., Bringas, J.,

Kells, A., Pivirotto, P., Bankiewicz, K., Vigneron, D.B., Nelson, S.J., 2014. Dynamic

hyperpolarized carbon-13 MR metabolic imaging of nonhuman primate brain. Magn.

Reson. Med. 71, 19–25.

Park, I., Larson, P.E.Z., Gordon, J.W., Carvajal, L., Chen, H.Y., Bok, R., Van Criekinge, M.,

Ferrone, M., Slater, J.B., Xu, D., Kurhanewicz, J., Vigneron, D.B., Chang, S.,

Nelson, S.J., 2018. Development of methods and feasibility of using hyperpolarized

carbon-13 imaging data for evaluating brain metabolism in patient studies. Magn.

Reson. Med. 80, 864–873.

Pellerin, Luc, Anne-Karine Bouzier-Sore, Aubert, Agnes, Serres, Sebastien, Michel Merle,

Costalat, Robert, Magistretti, P.J., 2007. Activity-dependent regulation of energy

metabolism by astrocytes: an update. Glia 55, 1416–1425.

P!erez-Escuredo, J., Van H!ee, V.F., Sboarina, M., Falces, J., Payen, V.L., Pellerin, L.,

Sonveaux, P., 2016. Monocarboxylate transporters in the brain and in cancer.

Biochim. Biophys. Acta Mol. Cell Res. 1863, 2481–2497.

Prichard, J., Rothman, D., Novotny, E., Petroff, O., Kuwabara, T., Avison, M.,

Howseman, A., Hanstock, C., Shulman, R., 1991. Lactate rise detected by 1

H NMR in

human visual cortex during physiologic stimulation. Proc. Natl. Acad. Sci. Unit. States

Am. 88, 5829–5831.

Schulte, R.F., Sacolick, L., Deppe, M.H., Janich, M.A., Schwaiger, M., Wild, J.M.,

Wiesinger, F., 2011. Transmit gain calibration for nonproton MR using the BlochSiegert shift. NMR Biomed. 24, 1068–1072.

Schulte, R.F., Sperl, J.I., Weidl, E., Menzel, M.I., Janich, M. a, Khegai, O., Durst, M.,

Ardenkjaer-Larsen, J.H., Glaser, S.J., Haase, A., Schwaiger, M., Wiesinger, F., 2013.

Saturation-recovery metabolic-exchange rate imaging with hyperpolarized [1-13C]

pyruvate using spectral-spatial excitation. Magn. Reson. Med. 69, 1209–1216.

J.T. Grist et al. NeuroImage 189 (2019) 171–179

178

極T代謝磁共振全球科研集錦

231

Sun, C.Y., Walker, C.M., Michel, K.A., Venkatesan, A.M., Lai, S.Y., Bankson, J.A., 2018.

Influence of parameter accuracy on pharmacokinetic analysis of hyperpolarized

pyruvate. Magn. Reson. Med. 79, 3239–3248.

Tomiyasu, M., Aida, N., Shibasaki, J., Tachibana, Y., Endo, M., Nozawa, K., Shimizu, E.,

Tsuji, H., Obata, T., 2016. Normal lactate concentration range in the neonatal brain.

Magn. Reson. Imaging 34, 1269–1273.

Wiesinger, F., Weidl, E., Menzel, M.I., Janich, M.A., Khegai, O., Glaser, S.J., Haase, A.,

Schwaiger, M., Schulte, R.F., 2012. IDEAL spiral CSI for dynamic metabolic MR

imaging of hyperpolarized [1-13C]pyruvate. Magn. Reson. Med. 68, 8–16.

Zaccagna, F., Grist, J.T., Deen, S.S., Woitek, R., Lechermann, L.M., McLean, M.A., Basu, B.,

Gallagher, F.A., 2018. Hyperpolarized carbon-13 magnetic resonance spectroscopic

imaging: a clinical tool for studying tumour metabolism. Br. J. Radiol. 91, 688.

J.T. Grist et al. NeuroImage 189 (2019) 171–179

179

極T代謝磁共振全球科研集錦

232

Imaging acute metabolic changes in mild traumatic brain injury patients using hyperpolarized [1-13C]pyruvate

研究背景

研究結(jié)果

研究對(duì)象

???????Traumatic brain injury, TBI?????????????代謝??????????????????

???????????????代謝????????????????????????代謝???????

????????????????????????????????????

????? TBI ????????????????????????????????????????????

?????????????????代謝???????????????????????????????

???????

CT ?磁共振?????????????????????????????????????????????

???????????研??????????????????????????極?] 1-13C]pyruvate 代謝?

??研??極???????????代謝???????????????????????代謝??????

2??????????TBI???

極T代謝磁共振全球科研集錦

233

研究結(jié)論

應(yīng)用方向

?極?] 1-13C]pyruvate ????????????????代謝??????????? CT?磁共振???????

????????????????????????????? [

13C] ???????????????? [13C] ??

????????代謝?????????????????????????? [

13C] ????????????

??????????????????????????????????????????????????

??????

?極?] 1-13C]pyruvate ????????代謝??????????????????????????????

??????????????????????????????????AD, PD ????極?] 1-13C]pyruvate

????????代謝??研??????????

???????????????????????????????????????????????????

????????代謝????

? ?????????代謝??研?

? ???????????????????????????????????

? ?????????????????????????????

????

????? TBI ???磁共振???????極?] 1-13C]pyruvate 代謝????? CT ?????????????

??????(A)(B)ǖ磁共振???????ǘ(C)(D)ǖ?????? ASL ??ǘ(E)(F)ǖ?極?] 1-13C]pyruvate ??ǘ(G)ǖ

?極?] 1-13C]pyruvate ??代謝????

????ǖ???35 ??88 ???168 ??????????2cm ????Glasgow ????15 ????????

?????????28 ???????T1W ? T2W FLAIR ?????????????????????? ASL

?????????????????????????????????????????極?] 1-13C]pyruvate 代

謝?? [

13C] ????????????20%?? [13C] ?????????????50%?

????ǖ???48 ??77 ???160 ?????????????6 ??????????Glasgow ????15 ??

?????2cm ????5cm ?????????6 ??????T1W ? T2W FLAIR ?????????????

???ASL ??????????????極?] 1-13C]pyruvate 代謝?? [

13C] ??????????? [13C] ???

??????????50%

極T代謝磁共振全球科研集錦

234

iScience

Article

Imaging Acute Metabolic Changes in Patients with

Mild Traumatic Brain Injury Using Hyperpolarized

[1-13C]Pyruvate

Edward P.

Hackett, Marco C.

Pinho, Crystal E.

Harrison, ...,

Surendra

Barshikar,

Christopher J.

Madden, Jae Mo

Park

jaemo.park@utsouthwestern.

edu

HIGHLIGHTS

Clinical translation of

hyperpolarized pyruvate

to TBI was demonstrated

Patients with mild TBI

were imaged with

hyperpolarized [1-13C]

pyruvate

Altered lactate and HCO3

–

production in the brain

nearest the site of trauma

Hackett et al., iScience 23,

101885

December 18, 2020 a 2020

The Author(s).

https://doi.org/10.1016/

j.isci.2020.101885

ll

OPEN ACCESS

極T代謝磁共振全球科研集錦

235

iScience

Article

Imaging Acute Metabolic Changes in Patients

with Mild Traumatic Brain Injury

Using Hyperpolarized [1-13C]Pyruvate

Edward P. Hackett,1 Marco C. Pinho,1,2 Crystal E. Harrison,1 Galen D. Reed,1,3 Jeff Liticker,1 Jaffar Raza,4

Ronald G. Hall,4 Craig R. Malloy,1,5 Surendra Barshikar,6 Christopher J. Madden,7 and Jae Mo Park1,2,8,9,

*

SUMMARY

Traumatic brain injury (TBI) involves complex secondary injury processes

following the primary injury. The secondary injury is often associated with rapid

metabolic shifts and impaired brain function immediately after the initial tissue

damage. Magnetic resonance spectroscopic imaging (MRSI) coupled with hyperpolarization of 13C-labeled substrates provides a unique opportunity to map the

metabolic changes in the brain after traumatic injury in real-time without invasive

procedures. In this report, we investigated two patients with acute mild TBI (Glasgow coma scale 15) but no anatomical brain injury or hemorrhage. Patients were

imaged with hyperpolarized [1-13C]pyruvate MRSI 1 or 6 days after head trauma.

Both patients showed significantly reduced bicarbonate (HCO3

–

) production, and

one showed hyperintense lactate production at the injured sites. This study reports the feasibility of imaging altered metabolism using hyperpolarized pyruvate in patients with TBI, demonstrating the translatability and sensitivity of

the technology to cerebral metabolic changes after mild TBI.

INTRODUCTION

Traumatic brain injury (TBI) causes mechanical damage and disruption of normal metabolism in the brain

(Corps et al., 2015). A major challenge for clinicians is managing complex secondary processes following

the primary injury. After the primary trauma, the surviving tissue undergoes metabolic shifts, resulting in

the development of potentially hazardous secondary metabolites and further damage. The secondary

events develop over a timescale after the primary injury, providing a potential window of opportunity

for detection and therapeutic intervention. Moreover, TBI is suspected to contribute to a variety of

chronic degenerative processes such as chronic traumatic encephalopathy, Alzheimer disease, and Parkinson disease (Smith et al., 2013). Numerous therapies have been proposed to reduce or prevent secondary brain damage, directly impacting long-term patient outcome (Xiong et al., 2015). Therefore,

the noninvasive detection and characterization of TBI pathophysiology during the acute and sub-acute

stages will have critical clinical implications and will be vital for identifying and developing effective

therapies.

Altered glucose metabolism and mitochondrial dysfunction are features of the pathophysiologic events

subsequent to TBI (Brooks and Martin, 2015; Kim et al., 2017). Invasive microdialysis methods have been

reported to study brain metabolism in these patients, but in spite of the critical importance of directly detecting mitochondrial function, no specific noninvasive methods exist. Positron emission tomography (PET)

detects uptake of radioactively labeled compounds such as glucose or acetate (e.g., [18F]FDG, [11C]acetate) in the human brain. Previous studies demonstrating cerebral hyperglycolysis in patients with TBI using

[

18F]fluorodeoxyglucose ([18F]FDG) PET measured cellular glucose uptake but could not measure the metabolic fate of glucose (Bergsneider et al., 1997). Another [18F]FDG-PET study that combined with microdialysis of patients with TBI reported that the rate of glucose metabolism correlates with microdialysate lactate

and pyruvate concentrations but not with the lactate-to-pyruvate ratio (Hutchinson et al., 2009), suggesting

an increase in glucose metabolism to both lactate and pyruvate, as opposed to a shift toward anaerobic

metabolism. Other studies indicated that the increased glucose uptake is directly related to the upregulated pentose phosphate pathway rather than to the rest of the glycolysis (Dusick et al., 2007).

1Advanced Imaging Research

Center, The University of

Texas Southwestern Medical

Center, Dallas, TX 75390, USA

2Department of Radiology,

The University of Texas

Southwestern Medical

Center, Dallas, TX 75390, USA

3GE Healthcare, Dallas, TX

75390, USA

4Department of Pharmacy

Practice, The Texas Tech

University Health Sciences

Center, Dallas, TX 75216, USA

5Department of Internal

Medicine, The University of

Texas Southwestern Medical

Center, Dallas, TX 75390, USA

6Department of Physical

Medicine & Rehabilitation,

The University of Texas

Southwestern Medical

Center, Dallas, TX 75390, USA

7Department of Neurological

Surgery, The University of

Texas Southwestern Medical

Center, Dallas, TX 75390, USA

8Department of Electrical and

Computer Engineering, The

University of Texas at Dallas,

Richardson TX 75080, USA

9Lead Contact

*Correspondence:

jaemo.park@utsouthwestern.

edu

https://doi.org/10.1016/j.isci.

2020.101885

iScience 23, 101885, December 18, 2020 a 2020 The Author(s).

This is an open access article under the CC BY-NC-ND license (http://creativecommons.org/licenses/by-nc-nd/4.0/). 1

ll

OPEN ACCESS

極T代謝磁共振全球科研集錦

236

Alternative way of imaging in vivo metabolism is using dynamic nuclear polarization (DNP) and rapid dissolution technique (Ardenkjaer-Larsen et al., 2003). Commercially available DNP polarizer can achieve more

than 100,000-fold signal amplification of magnetic resonance-detectable (e.g., 13C-labeled) substrates in

liquid state. Administration of hyperpolarized 13C-labeled substrates in coordination with 13C magnetic

resonance spectroscopy imaging (MRSI) opened new opportunities to assess in vivo metabolic processes

of individual enzyme-catalyzed reactions in various organs including the brain. In particular, previous animal studies using hyperpolarized [1-13C]pyruvate demonstrated that increased [1-13C]lactate production in

the injured brain tissue was associated with microglial activation (DeVience et al., 2017; Guglielmetti et al.,

2017). Metabolites that are typically detectable in the cerebral 13C spectrum using hyperpolarized [1-13C]

pyruvate are [1-13C]lactate and [13C]bicarbonate (HCO3

–

) as shown in Figure 1. The yellow arrows indicate

metabolic shifts in acute TBI (DeVience et al., 2017). In this study, we translated these preclinical discoveries

to demonstrate metabolic abnormalities in the brain of patients after acute mild TBI.

RESULTS

Two patients with acute TBI were recruited from the Parkland Health and Hospital System Emergency

Room. Computed tomographic (CT) scans at admission confirmed that neither patient had underlying fractures, hemorrhage, or other anatomical injury. Patient #1 was a daily smoker, and Patient #2 was diabetic.

Otherwise, both subjects were generally healthy and had no history of mental illness or alcoholism. Both

patients tolerated the examination well.

The first patient was a 35-year-old African American male (88 kg, 168 cm) with a 2-cm laceration to the left

frontal scalp due to blunt force trauma (whipped by a metal gun) with a Glasgow Coma Scale (GCS) score

of 15 and no loss of consciousness (LOC). Figure 2 shows axial images of this patient 28 h after the head

trauma. Besides left frontal scalp cutaneous edema/ecchymosis, no cerebral contusion or anatomical

lactate

glucose

glycolysis

cytoplasm

mitochondrion

TCA cycle

1 2 3

[1-13C]pyruvate pyruvate

acetyl-CoA

PDH

PC

LDH

CO2

CO2

HCO3

-

13C

12C

citrate

fumarate

malate

PEP OAA

α-ketoglutarate

CO2

CO2

Figure 1. Metabolic Fate of Hyperpolarized [1-13C]Pyruvate in the Brain

[1-13C]Pyruvate (black circle: 13C, white circle: 12C) is converted either to [1-13C]lactate via LDH or acetyl-CoA and 13CO2

(and [13C]HCO3

-

) through PDH. Possibly, [1-13C]pyruvate can be also converted into [1-13C]oxaloacetate (OAA) through

PC, eventually releasing the labeled carbon (13C) as CO2. The gray circle indicates 13C after backward scrambling of OAA

to fumarate and malate in the TCA cycle. PEP: phosphoenolpyruvate; LDH: lactate dehydrogenase; PDH: pyruvate

dehydrogenase; PC: pyruvate carboxylase.

ll

OPEN ACCESS

2 iScience 23, 101885, December 18, 2020

iScience

Article

極T代謝磁共振全球科研集錦

237

brain damage was identified in the T1-weighted and T2-weighted fluid-attenuated inversion recovery

(FLAIR) images. Arterial spin labeling (ASL) detected a hyperintense region at the site of the left frontal

scalp injury but unremarkable appearance of cerebral perfusion. Likewise, total hyperpolarized 13C signal

was hyperintense in the injured scalp (Figure 2D) but comparable between the left and right brain hemispheres. In contrast, [13C]HCO3

– and [1-13C]lactate production from hyperpolarized [1-13C]pyruvate was

altered in the injured side of the brain (left frontal lobe, underlying the scalp injury) when compared

with the contralateral side (Figures 2E and 2F). [13C]HCO3

– signal, averaged over the injured region of

interest and normalized with total 13C (TC) signal (ROITBI, HCO3

–

/TC = 0.027), was 54.0% lower than

that in the contralateral ROI (ROICTL, HCO3

–

/TC = 0.059). Conversely, lactate production in the injured

region (lactate/TC = 0.236) was larger by 23.7% than the unimpacted contralateral brain region

(lactate/TC = 0.191).

The second patient (Patient #2; 48 years old, 77 kg, 160 cm, Hispanic, male) sustained a head injury as a

consequence of an !6-m fall from a construction site scaffolding (GCS 15, 2 min LOC). When recruited,

he had a 2-cm laceration to the left medial eyebrow and 5-cm subcutaneous hematoma extending laterally

from the laceration. He was studied using the same 1

H/13C magnetic resonance imaging (MRI) protocol

6 days after the initial injury. Similar to Patient #1, no structural brain damage was found in the 1H MRI,

but smaller [13C]HCO3

– was observed in the injured region (HCO3–

/TC = 0.025) than the contralateral

side of the brain (0.047) (Figure 3). [1-13C]Lactate level was comparable (lactate/TC = 0.164 for the injured

region and 0.162 for the contralateral region).

A B

G

C D

E F

Figure 2. Metabolism of Acute TBI in Patient #1 Imaged by Hyperpolarized 13C Pyruvate

The patient (35 years old, male), injured on the left frontal scalp (GCS 15, no LOC), was imaged 28 h after the injury.

(A) An axial slice that includes the injured region was prescribed for 13C imaging.

(B and C) (B) T1-weighted and T2-weighted 1

H FLAIR and (C) ASL images showed swelling and increased perfusion in the

injured site outside the skull (yellow circle), whereas no cerebral contusion, hemorrhage, or hypoperfusion was detected.

(D) Region with increased total hyperpolarized 13C (TC) signal image matched to the scalp edema.

(E and F) (E) Increased [1-13C]lactate conversion and (F) decreased [13C]HCO3

– production were detected in the brain

tissues of the injured hemisphere.

(G) Averaged spectra over the injured brain region (ROITBI, dotted spectra) and the contralateral normal-appearing brain

region (ROICTL, solid spectra).

ll

OPEN ACCESS

iScience 23, 101885, December 18, 2020 3

iScience

Article

極T代謝磁共振全球科研集錦

238

DISCUSSION

A common feature reported in patients with TBI is the increase of glucose consumption rate with no parallel

increase in mitochondrial oxidative phosphorylation, known as hyperglycolysis (Bergsneider et al., 1997).

Hyperglycolysis coupled with mitochondrial dysfunction should favor metabolism of pyruvate, the end

product of glycolysis, to lactate via lactate dehydrogenase, rather than oxidation to acetyl-CoA via pyruvate dehydrogenase (PDH). The hyperintense [1-13C]lactate signal in the impacted region from Patient

#1 is consistent with previous in vivo preclinical imaging studies. DeVience et al. showed significant increase of [1-13C]lactate production from hyperpolarized [1-13C]pyruvate 4 h post-injury using a controlled

cortical impact rat model (DeVience et al., 2017), and Guglielmetti et al. reported longitudinal changes of

lactate production using this technique (Guglielmetti et al., 2017). The absence of hyperglycolysis in Patient

#2 could be due to the severity of the injury (Bergsneider et al., 1997) or metabolic transition from hyperglycolysis to hypometabolic stage (Greco et al., 2020; Jalloh et al., 2015).

Mitochondrial dysfunction plays a key role in the pathophysiology of TBI (Kim et al., 2017). [13C]HCO3

– production from hyperpolarized [1-13C]pyruvate reflects PDH flux. As PDH is an enzyme complex that is integrated into the inner membrane of the mitochondria, the decreased [13C]HCO3

– production in the injured

brain region implies mitochondrial injury/dysfunction in spite of normal anatomy, which is consistent with

previous animal study (DeVience et al., 2017). It should be noted that [13C]HCO3

– can be also produced via

an anaplerotic pathway into the tricarboxylic acid cycle, which is exclusively achieved by pyruvate carboxylase (PC), an astrocyte-specific enzyme (Shank et al., 1985), as shown in Figure 1. Pyruvate carboxylation in

A B

G

C D

E F

Figure 3. Hyperpolarized 13C Pyruvate Imaging of Patient #2

The patient had a 2-cm laceration to the left medial eyebrow and 5-cm hematoma extending laterally from the laceration

from a head injury (GCS 15, 2 min LOC). Images were acquired 6 days from the injury.

(A–C) (A) 13C slice was prescribed to include the laceration. Axial 1

H images of (B) T2-weighted FLAIR and (C) ASL showed

the injured region outside the skull (yellow circle).

(D) Besides the scalp hematoma, no abnormal distribution of total hyperpolarized 13C signals (TC).

(E and F) (E) Lactate was comparable between the impacted region and the contralateral side of the brain, whereas (F)

decreased HCO3

– signal was observed in the impacted brain region.

(G) Averaged spectra over the injured brain region (ROITBI, dotted spectra) and the contralateral normal-appearing brain

region (ROICTL, solid spectra).

ll

OPEN ACCESS

4 iScience 23, 101885, December 18, 2020

iScience

Article

極T代謝磁共振全球科研集錦

239

astrocyte increases during the secondary injury process, whereas pyruvate utilization via PDH is downregulated to protect neurons (Bartnik-Olson et al., 2013).

As shown in the Figures 2D and 3D, hyperpolarized 13C signals were predominant in the gray matter rather

than the white matter or subcutaneous tissues. These are consistent with previous human brain studies

using hyperpolarized pyruvate (Gordon et al., 2019; Lee et al., 2019). Besides the directly impacted brain

regions, the lactate and HCO3

– maps showed larger variation in the patients when compared with those

acquired from healthy subjects in previous studies. For instance, reduced HCO3

– production was observed

in the right posterior brain region of Patient #1, indicating possible contrecoup injury in the region.

Both patients were overweight. Patient #1 was very muscular with body mass index (BMI) of 31.2 but did not

have a diagnosis of type 2 diabetes and was not on any hypoglycemic medications. Patient #2 (BMI = 30.1)

did have a diagnosis of type 2 diabetes on metformin (500 mg orally twice daily). Plasma glucose levels of

the patients were well within 160 mg/dL (108 for patient #1, 133 for patient #2), which is representative of

normal brain biochemistry for clinical purposes as measured by PET (Varrone et al., 2009; Viglianti et al.,

2017). Although diffuse changes in pyruvate metabolism could be postulated due to hyperglycemia, it

seems unlikely that regional abnormalities as observed here would occur.

The specific timing of this glucose dysmetabolism is currently not known. Longitudinal monitoring with a

larger patient population will be needed to further evaluate the utility of hyperpolarized pyruvate for

noninvasive assessment of the TBI metabolism. If this result can be verified, it would have a major impact

on the evaluation of patients involved in sports, military activities, or those who suffer from assault or accidents. Considering that this technique is safe and well-tolerated by patients, this technology may prove

valuable for managing patients experiencing secondary injury processes, identifying early treatment

response, and accelerating drug development efforts. In fact, pyruvate has been suggested as a

neuro-protective substrate with therapeutic effect (Moro et al., 2016). Patients with moderate and severe

TBI are expected to have larger metabolic alteration with potential challenges of imaging data interpretation due to morphological distortion and hemorrhage. Beyond acute and sub-acute TBI, the imaging

methods and associated biomarkers to be developed under this application will be likely also applicable

to concussion.

In this study, we investigated cerebral metabolism in patients with acute TBI using hyperpolarized [1-13C]

pyruvate. We found altered cerebral metabolism in the brain nearest the site of trauma despite no visible

anatomical damage in the brain on MRI, demonstrating the sensitivity of hyperpolarized pyruvate to

altered metabolism in TBI. The acute metabolic changes in [1-13C]lactate and [13C]HCO3

– images in the

injured region indicate the potential clinical utility of hyperpolarized [1-13C]pyruvate in managing patients

with TBI and provides objective evidence of injury even when conventional studies with CT and MRI are

unrevealing.

Limitations of the Study

This study demonstrated the feasibility of hyperpolarized [1-13C]pyruvate for imaging metabolic changes

following a mild brain injury in humans. Although we could detect clear alteration of pyruvate metabolism

in the injured brain, biological interpretation of the imaging biomarker needs to be careful and requires

further clarification. First, production of [1-13C]lactate signal reflects both metabolic flux from pyruvate

to lactate and isotopic chemical exchange between pyruvate and lactate (Kennedy et al., 2012). The latter

can be significantly affected by the intrinsic lactate pool size (Hurd et al., 2013). Second, unlike [13C]HCO3

–

,

[1-13C]lactate signal detected in the brain is not exclusively produced by brain tissues as [1-13C]lactate produced in the vasculature or other organs (Wespi et al., 2018; Xu et al., 2011) can be delivered to the brain.

Moreover, although lactate is a preferred energy substrate for neurons (Be′ langer et al., 2011; Magistretti

and Allaman, 2018), it is likely that the excessive cerebral presence of hyperpolarized pyruvate results in

cellular transport of pyruvate (and lactate) into both neurons and glia via monocarboxylate transporters.

The large spatial resolution is another limitation of the study. Although the large voxel size was helpful

to achieve reliable detection of HCO3

-

, it was unavoidable to experience partial volume effects in the 13C images. Imaging with improved spatial resolution will be required for accurate regional assessment

of TBI metabolism and to reduce the partial volume effects. The statistical analysis was not performed in

this pilot study primarily due to the small number of subjects and the difficulty of patient recruitments in

such an acute stage. Considering the diversity of brain injury types and the complexity of secondary

ll

OPEN ACCESS

iScience 23, 101885, December 18, 2020 5

iScience

Article

極T代謝磁共振全球科研集錦

240

injuries, studies with a larger number of patients with longitudinal follow-ups will be necessary for further

evaluation of the imaging technique and for systematic characterization of the metabolic alterations.

Resource Availability

Lead Contact

Further information and requests for resources should be directed to and will be fulfilled by the Lead Contact, Jae Mo Park (jaemo.park@utsouthwestern.edu).

Materials Availability

This study did not generate new unique reagents or materials.

Data and Code Availability

Original/source data or images in the paper is available upon request.

METHODS

All methods can be found in the accompanying Transparent Methods supplemental file.

SUPPLEMENTAL INFORMATION

Supplemental Information can be found online at https://doi.org/10.1016/j.isci.2020.101885.

ACKNOWLEDGMENTS

Funding: The Texas Institute for Brain Injury and Repair of Peter O’Donnell Jr. Brain Institute; The Mobility

Foundation; National Institutes of Health of the United States (1R01NS107409-01A1, 5P41EB015908-32,

1S10OD018468-01); The Welch Foundation (I-2009-20190330). Personnel Support: We appreciate the

research nurses and the MR technicians of the Advanced Imaging Research Center at UT Southwestern—Lucy Christie, Jeannie Baxter, Kelley Derner, Maida Tai, and Salvador Pena.

AUTHOR CONTRIBUTIONS

M.C.P., C.J.M., and J.M.P. designed research; E.P.H., M.C.P., S.B., C.J.M., and J.M.P. recruited patients;

E.P.H., C.E.H., J.L., J.R., G.D.R., C.R.M., and J.M.P. performed research; E.P.H., M.C.P., C.J.M., and

J.M.P. analyzed data; E.P.H., M.C.P., C.R.M., C.J.M., and J.M.P. wrote the paper.

DECLARATION OF INTERESTS

G.D.R. is an employee of GE Healthcare.

Received: July 20, 2020

Revised: October 25, 2020

Accepted: November 25, 2020

Published: December 18, 2020

REFERENCES

Ardenkjaer-Larsen, J.H., Fridlund, B., Gram, A.,

Hansson, G., Hansson, L., Lerche, M.H., Servin, R.,

Thaning, M., and Golman, K. (2003). Increase in

signal-to-noise ratio of > 10,000 times in liquidstate NMR. Proc. Natl. Acad. Sci. U S A 100,

10158–10163.

Bartnik-Olson, B.L., Harris, N.G., Shijo, K., and

Sutton, R.L. (2013). Insights into the metabolic

response to traumatic brain injury as revealed by

(13)C NMR spectroscopy. Front. Neuroenergetics

5, 8.

Bergsneider, M., Hovda, D.A., Shalmon, E., Kelly,

D.F., Vespa, P.M., Martin, N.A., Phelps, M.E.,

McArthur, D.L., Caron, M.J., Kraus, J.F., and

Becker, D.P. (1997). Cerebral hyperglycolysis

following severe traumatic brain injury in humans:

a positron emission tomography study.

J. Neurosurg. 86, 241–251.

Be′ langer, M., Allaman, I., and Magistretti, P.J.

(2011). Brain energy metabolism: focus on

astrocyte-neuron metabolic cooperation. Cell

Metab. 14, 724–738.

Brooks, G.A., and Martin, N.A. (2015). Cerebral

metabolism following traumatic brain injury: new

discoveries with implications for treatment. Front.

Neurosci. 8, 408.

Corps, K.N., Roth, T.L., and McGavern, D.B.

(2015). Inflammation and neuroprotection in

traumatic brain injury. JAMA Neurol. 72, 355–362.

DeVience, S.J., Lu, X., Proctor, J., Rangghran, P.,

Melhem, E.R., Gullapalli, R., Fiskum, G.M., and

Mayer, D. (2017). Metabolic imaging of energy

metabolism in traumatic brain injury using

hyperpolarized [1-13C]pyruvate. Sci. Rep. 7, 1907.

Dusick, J.R., Glenn, T.C., Lee, W.N.P., Vespa,

P.M., Kelly, D.F., Lee, S.M., Hovda, D.A., and

Martin, N.A. (2007). Increased pentose

phosphate pathway flux after clinical traumatic

brain injury: a [1,2-13C2]glucose labeling study in

humans. J. Cereb. Blood Flow Metab. 27, 1593–

1602.

Gordon, J.W., Chen, H.-Y., Autry, A., Park, I., Van

Criekinge, M., Mammoli, D., Milshteyn, E., Bok,

R., Xu, D., Li, Y., et al. (2019). Translation of

ll

OPEN ACCESS

6 iScience 23, 101885, December 18, 2020

iScience

Article

極T代謝磁共振全球科研集錦

241

Carbon-13 EPI for hyperpolarized MR molecular

imaging of prostate and brain cancer patients.

Magn. Reson. Med. 81, 2702–2709.

Greco, T., Vespa, P.M., and Prins, M.L. (2020).

Alternative substrate metabolism depends on

cerebral metabolic state following traumatic

brain injury. Exp. Neurol. 329, 113289.

Guglielmetti, C., Chou, A., Krukowski, K., Najac,

C., Feng, X., Riparip, L.-K., Rosi, S., and Chaumeil,

M.M. (2017). In vivo metabolic imaging of

Traumatic Brain Injury. Sci. Rep. 7, 17525.

Hurd, R.E., Spielman, D., Josan, S., Yen, Y.-F.,

Pfefferbaum, A., and Mayer, D. (2013). Exchangelinked dissolution agents in dissolution-DNP (13)

C metabolic imaging. Maregn Reson. Med. 70,

936–942.

Hutchinson, P.J., O’Connell, M.T., Seal, A.,

Nortje, J., Timofeev, I., Al-Rawi, P.G., Coles, J.P.,

Fryer, T.D., Menon, D.K., Pickard, J.D., and

Carpenter, K.L.H. (2009). A combined

microdialysis and FDG-PET study of glucose

metabolism in head injury. Acta Neurochir (Wien)

151, 51–61, discussion 61.

Jalloh, I., Carpenter, K.L.H., Helmy, A., Carpenter,

T.A., Menon, D.K., and Hutchinson, P.J. (2015).

Glucose metabolism following human traumatic

brain injury: methods of assessment and

pathophysiological findings. Metab. Brain Dis. 30,

615–632.

Kennedy, B.W.C., Kettunen, M.I., Hu, D.-E., and

Brindle, K.M. (2012). Probing lactate

dehydrogenase activity in tumors by measuring

hydrogen/deuterium exchange in hyperpolarized

l-[1-(13)C,U-(2)H]lactate. J. Am. Chem. Soc. 134,

4969–4977.

Kim, S., Han, S.C., Gallan, A.J., and Hayes, J.P.

(2017). Neurometabolic indicators of

mitochondrial dysfunction in repetitive mild

traumatic brain injury. Concussion 2, CNC48.

Lee, C.Y., Soliman, H., Geraghty, B.J., Chen, A.P.,

Connelly, K.A., Endre, R., Perks, W.J., Heyn, C.,

Black, S.E., and Cunningham, C.H. (2019). Lactate

topography of the human brain using

hyperpolarized 13C-MRI. Neuroimage 204,

116202.

Magistretti, P.J., and Allaman, I. (2018). Lactate in

the brain: from metabolic end-product to

signalling molecule. Nat. Rev. Neurosci. 19,

235–249.

Moro, N., Ghavim, S.S., Harris, N.G., Hovda, D.A.,

and Sutton, R.L. (2016). Pyruvate treatment

attenuates cerebral metabolic depression and

neuronal loss after experimental traumatic brain

injury. Brain Res. 1642, 270–277.

Shank, R.P., Bennett, G.S., Freytag, S.O., and

Campbell, G.L. (1985). Pyruvate carboxylase: an

astrocyte-specific enzyme implicated in the

replenishment of amino acid neurotransmitter

pools. Brain Res. 329, 364–367.

Smith, D.H., Johnson, V.E., and Stewart, W.

(2013). Chronic neuropathologies of single and

repetitive TBI: substrates of dementia? Nat. Rev.

Neurol. 9, 211–221.

Varrone, A., Asenbaum, S., Vander Borght, T.,

Booij, J., Nobili, F., Na? gren, K., Darcourt, J.,

Kapucu, O.L., Tatsch, K., Bartenstein, P., and Van

Laere, K.; European Association of Nuclear

Medicine Neuroimaging Committee (2009).

EANM procedure guidelines for PET brain

imaging using [18F]FDG, version 2. Eur. J. Nucl.

Med. Mol. Imaging 36, 2103–2110.

Viglianti, B.L., Wong, K.K., Wimer, S.M.,

Parameswaran, A., Nan, B., Ky, C., Townsend,

D.M., Rubello, D., Frey, K.A., and Gross, M.D.

(2017). Effect of hyperglycemia on brain and liver

18F-FDG standardized uptake value (FDG SUV)

measured by quantitative positron emission

tomography (PET) imaging. Biomed.

Pharmacother. 88, 1038–1045.

Wespi, P., Steinhauser, J., Kwiatkowski, G., and

Kozerke, S. (2018). Overestimation of cardiac

lactate production caused by liver metabolism of

hyperpolarized [1-13 C]pyruvate. Magn. Reson.

Med. 80, 1882–1890.

Xiong, Y., Zhang, Y., Mahmood, A., and Chopp,

M. (2015). Investigational agents for treatment of

traumatic brain injury. Expert Opin. Investig.

Drugs 24, 743–760.

Xu, T., Mayer, D., Gu, M., Yen, Y.-F., Josan, S.,

Tropp, J., Pfefferbaum, A., Hurd, R., and

Spielman, D. (2011). Quantification of in vivo

metabolic kinetics of hyperpolarized pyruvate in

rat kidneys using dynamic 13C MRSI. NMR

Biomed. 24, 997–1005.

ll

OPEN ACCESS

iScience 23, 101885, December 18, 2020 7

iScience

Article

極T代謝磁共振全球科研集錦

242

iScience, Volume 23

Supplemental Information

Imaging Acute Metabolic Changes in Patients

with Mild Traumatic Brain Injury

Using Hyperpolarized [1-13C]Pyruvate

Edward P. Hackett, Marco C. Pinho, Crystal E. Harrison, Galen D. Reed, Jeff

Liticker, Jaffar Raza, Ronald G. Hall, Craig R. Malloy, Surendra Barshikar, Christopher

J. Madden, and Jae Mo Park

極T代謝磁共振全球科研集錦

243



Supplementary Information

Transparent Methods

Human subjects and patient recruitment criteria

This study was approved by the local Institutional Review Board (IRB) of the (niversity of Texas

Southwestern Medical Center (IRB: ST( 072017-009; ClinicalTrials.gov NCT03502967). A written and

informed consent was obtained from all the study participants. The imaging protocol was fully compliant

with HIPPA regulation. Patients between the age of 18-60 years old in good general health were screened

for enrollment if they had sustained a mild injury to the head with no structural brain damage detected in

any CT scans within 10 days of trauma. Patients were further screened to exclude any confounding

comorbidities including a history of mental illness, history of substance abuse, and other major medical

conditions. Patients with known prior cardiovascular or neurological disease were excluded.

1

H/13C MRI protocol

All the studies were performed using a clinical SPINlab

polarizer (GE Healthcare, *aukesha,

*I (SA), a 3T wide-bore MR scanner (GE Healthcare, 750w Discovery), and a 13C/1

H dual-frequency (

1

H:

quadrature transmit/receive, 13C: quadrature transmit/8-channel phased array receive) nested-design

radiofrequency (RF) head coil (Clinical MR Solutions, Brookfield, *I (SA) (Ma et al., 2019). Both study

participants in this report were recruited at the Parkland Health and Hospital System Emergency

Department, a (T Southwestern-affiliated hospital in Dallas. *hen admitted, a full workup that includes

primary and secondary trauma survey, blood tests, and CT imaging of head and C-spine was initiated and

the CT scans showed no evidence of intracranial injury for either patient. The patients were released with

minor laceration care and brought in for research imaging. The subjects were imaged with a brain MR

protocol, which includes an injection of hyperpolarized [1-

13C]pyruvate (IND: 133229). The

hyperpolarized pyruvate solution was injected after a two-dimensional T2-weighted fluid attenuated

inversion recovery (FLAIR) scan. For 13C acquisition, a volume of 250-mM hyperpolarized pyruvate

極T代謝磁共振全球科研集錦

244

2

corresponding to a 0.1 mmol/kg dose was injected, followed by a 25-mL saline flush. The injection rate

was 5 mL/sec. A single-slice two-dimensional (2D) dynamic spiral chemical shift imaging (spiral CSI) was

used for dynamic imaging of hyperpolarized 13C signals (Park et al., 2012; 2019).

Each subject was scanned by a 3-plane 1

H gradient echo sequence (fast GRE; field of view [FOV]

= 32 cm, echo time [TE] = 1.1 msec, repetition time [TR] = 3.6 msec, slice thickness = 10 mm), followed

by a 2D axial T2-weighted fluid-attenuated and inversion recovery (T2w FLAIR; FOV = 24 cm ′ 24 cm,

TE = 144.7 ms, TR = 8 sec, inversion time [TI] = 2163 msec, slice thickness = 5 mm, number of slices =

16, flip-angle [FA] = 160o) as a basic assessment of axonal/parenchymal injuries that are typical

manifestations of trauma. Prior to 13C imaging, B0 inhomogeneity of the target 13C slice was minimized

using a single-voxel 1

H point-resolved spectroscopy (PRESS) sequence by adjusting shim gradients up to

1st order. For 13C metabolic imaging, 13C spiral chemical shift imaging (FOV = 24 cm ′ 24 cm, matrix size

= 16 ′ 16, slice thickness = 3 cm, variable FA up to 30o per timepoint, TR = 5 sec, 7 spatial interleaves of

spiral readout, spectral width = 814 Hz, 48 echoes) was used with a bolus injection of hyperpolarized [1-

13C]pyruvate. The 13C scan was initiated 5-sec after the start of injection. The 13C transmit power was precalibrated with a gadolinium-doped 0.4-M spherical [13C]HCO3

– phantom (diameter = 18 cm).After the 13C

acquisition, 1

H images were further acquired from each subject for mapping brain anatomy: T1-weighted

FLAIR (FOV = 24 cm ′ 24 cm, TE = 9.4 msec, TR = 3109 msec, inversion time [TI] = 1215 msec, slice

thickness = 5 mm, number of slices = 16, FA = 142o

), a 3D pseudo-continuous arterial spin labeling (ASL)

sequence with stack of spiral readouts (FOV = 24 cm ′ 24 cm ′ 16.2 cm, TE = 9.9 msec, TR = 6733 msec,

post-labeling delay = 2025 msec, 3 averages, matrix size = 512/interleave, 8 interleaves, flip angle = 90o

).

ASL was performed to identify vascular imbalance of the injured region (e.g., hypoperfusion).

Reconstruction of 13C data

All 13C data were reconstructed using MATLAB (Mathworks, Natick, MA). After apodization by

a 10-Hz Gaussian filter, the raw data were zero-filled in spectral and spatial domains by a factor of 4,