| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
Tumor Biology |
Instituto de Investigaciones Biomédicas, Consejo Superior de Investigaciones Cientificas, 28029 Madrid, Spain [M-L. G-M., S.C.]; Unité mixte Institut National de la Santé et de la Recherche Médicale/Université Joseph Fourier: U438 "RMN Bioclinique," Laboratoire de Recherche Correspondant du Commissariat à l’Energie Atomique, Centre Hospitalier Universitaire BP 217, 38043 Grenoble, France [G. H., C. R., R. F., J. A. C., A. Z.]; and Departemento de Química Orgánica y Biología, Facultad de Ciencias, Universidad Nacional de Educación a Distancìa, 28040 Madrid, Spain [P. B.]
 |
ABSTRACT |
|---|
 Top ABSTRACT INTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONREFERENCES |
|---|
10 mm in diameter. Three mmols of IEPA were
injected in the right jugular vein from t = 0 to t = 60 min. From
t = 50 min to t = 90
min, spin-echo 1H SI was performed with an echo time of 40
ms in a 2.5-mm slice including the glioma (nominal voxel size, 2.2
µl). IEPA resonances were detected only within the glioma and were
intense enough for pHe to be calculated from the chemical
shift of the H2 resonance in almost all voxels of the glioma.
1H spectroscopic images with an echo time of 136 ms were
then acquired to map metabolites: lactate, choline-containing compounds
(tCho), phosphocreatine/creatine, and N-acetylaspartate.
Finally, T1-weighted imaging after injection of a bolus of
Gd-DOTA gave a map indicative of extravasation. On average, the
gradient of pHe (measured where sufficient IEPA was
present) from the center to the periphery was not statistically
significant. Mean pHe was calculated for each of the four
gliomas, and the average was 7.084 ± 0.017 (± SE;
n = 4 rats), which is acid with respect
to pHe of normal tissue. After normalization of spectra to
their water peak, voxel-by-voxel comparisons of peak areas showed that
N-acetylaspartate, a marker of neurons, correlated negatively with IEPA
(P < 0.0001) and lactate
(P < 0.05), as expected of a glioma
surrounded by normal tissue. tCho (which may indicate
proliferation) correlated positively with pHe
(P < 0.0001). Lactate correlated
positively with tCho (P < 0.0001),
phosphocreatine/creatine (P < 0.001),
and Gd-DOTA (P < 0.0001). Although
lactate is exported from cells in association with protons, within the
gliomas, no evidence was observed that pHe was
significantly lower where lactate concentration was higher. These
results suggest that lactate is produced mainly in viable,
well-perfused, tumoral tissue from which proton equivalents are rapidly
cleared. |
INTRODUCTION |
|---|
|
TopABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONREFERENCES |
|---|
Tumor pHe has been measured mainly by invasive techniques that measure it at a single point, namely microelectrodes (9 , 10) and miniature optical probes (11) . The value of these measurements depends on how uniform pHe is throughout the tumor. Mean pHe within tumors has also been measured noninvasively by NMR using extracellular probe molecules containing 31P or 19F (6 , 12, 13, 14) . Spatial variations in pHe have been measured over distances in the order of 100 µm in the tissue between blood vessels in the exposed superficial layers of a s.c. tumor by optical techniques (15) . In the present work, we have made maps of pHe on a larger scale throughout sections of C6 gliomas in rat brain. The tool we used is a new probe molecule, which has pH-dependent 1H resonances detectable by 1H NMR spectroscopy. This molecule is IEPA. It has been shown that IEPA does not enter erythrocytes (compound 9 in Ref. 16 ), and it appears to remain extracellular in a tumor model in a mouse mammary fat pad where the 1H signal was sufficient for SI (17) . Preliminary results had shown that systemically delivered IEPA infiltrates the extracellular space of C6 gliomas in brain and allows mapping of pHe (18) .
In addition to providing a signal:noise ratio sufficient to allow imaging of pHe rather than just measurement of the average value in a volume including the tumor, detection by 1H magnetic resonance spectroscopy has an additional advantage: using the same radiofrequency coil, the distributions of various endogenous compounds with 1H resonances can be readily imaged in the same experiments. These include compounds of which the concentrations might be causally related to the value of pHe, notably lactate. Hence, in this paper, we not only report the use of IEPA to image the distribution of pHe within C6 gliomas, but we have compared this distribution with the distribution of lactate and also of tCho, NAA, and tCr. The results lead us to consider the reasons why pHe in C6 gliomas should be acid. Both normal brain tissue (19) and tumor cells in particular (20 , 21) produce lactate even under aerobic conditions. Lactate is exported from cells in association with H+ (22) and in this way is expected to contribute to the extracellular acidity. However, Yamagata et al. (23) and Newell et al. (24) have found that even tumor cells lacking lactate dehydrogenase, which produced very little lactate in vitro, nevertheless created an acid extracellular environment when grown as tumors. By using IEPA, we have been able to see whether the distribution of pHe correlated with that of lactate in vivo.
|
MATERIALS AND METHODS |
|---|
|
TopABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONREFERENCES |
|---|
Calibration of the pH Dependence of the Chemical Shift of the
IEPA H2 Resonance.
Fifteen vials containing solutions of 20 mM IEPA and 100
mM of the HEPES buffer in rat plasma were titered at 37°C
with HCl or NaOH to pHs in the range 4.5–8.5. The chemical shift of
the IEPA H2 hydrogen and its T2 were measured at
37°C using an 8.4 Tesla magnet (Oxford Instruments, United Kingdom)
interfaced with an AM-360 NMR spectrometer (Bruker, Karlsruhe,
Germany). We used a single pulse sequence with a 1 s presaturating
pulse on the water resonance to decrease the water resonance intensity.
The excitation pulse duration was 7 µs, data size 16384 points, and
acquisition time 0.95 s. pH titrations of the H2 chemical shift
were fitted using Sigmaplot v 4.0 (SPSS Inc., Chicago, IL) with the
Henderson-Hasselbalch equation:
 | (1) |
is the chemical shift and
1 >
2 are the asymptotic values. The T2 of IEPA H2 and H5 peaks were also determined in the same samples used for the pH titration. The Hahn spin-echo sequence was modified to include a 1-s selective presaturating pulse, a jump and return excitation pulse (interpulse delay, 210 µs), and a binomial refocusing pulse (interpulse delay, 420 µs; Ref. 25 ). Ten different values of TE in the range 2–400 ms were used to obtain T2 at different pHs. T2(H2) showed a minimum of 43 ms at pH 6.5 increasing to 81 ms at pH 4.5 and to 100 ms at pH 8.5. Although this variation in T2 does not affect the calculation of pH from the chemical shift, it would affect the apparent distribution of IEPA. Therefore, the distribution of IEPA was calculated from the H5 peak of which the T2 did not vary significantly over the pH range 5.5–7.5 (T2 = 57 ± 4 ms).
Animal Preparation.
All of the procedures involving animals conformed to the guidelines of
the French Government (decree 87–848, October 19, 1987, licenses
006683 and A38071). To prepare the glioma model, Wistar rats (200–230
g) were anesthetized with chloralhydrate (400 mg/kg), and
105 C6 glioma cells (26)
in DMEM
were injected stereotaxically in the right caudate nucleus (as in Refs.
27
, 28 but without the use of agar). Gliomas developed,
which, after 3 weeks, occupied 30–50% of the right hemisphere of the
brain. All of the rats were females except for one male included in the
results shown in Fig. 4
.
|
15 mmol/kg body
weight. Another 0.7-mm indwelling catheter was inserted into the left
femoral artery. Blood gases (PaO2and PaCO2),
arterial oxygen saturation of hemoglobin
(SaO2),
pHa, and hemoglobin content were measured in
arterial blood samples of <0.1 ml (ABL 510; Radiometer, Copenhagen,
Denmark) before the beginning of the IEPA infusion and subsequently at
40 min intervals up to 2 h (Fig. 1)
. Rectal temperature was maintained at 36.5–37.5°C by a heating pad
placed on the abdomen. Measurements of blood gases and pHawere corrected for rectal temperature. pHawas in the range 7.28–7.45, and during the NMR imaging,
arterial carbon dioxide pressure was in the range 32–40 mm of
Hg. At the end of the cycle of measurements, the animal was killed by
an overdose of chloralhydrate.
|
The rat was prone, its head secured by ear bars, and a 25-mm diameter
surface coil located directly above the brain was used for
radiofrequency transmission and reception. After radiofrequency coil
matching and tuning, the magnetic field homogeneity was coarsely
adjusted to obtain a line width for water <40 Hz in a 5-mm horizontal
slice of the brain. Additional fine adjustments were carried out in the
volume of interest for each separate NMR measurement. An idealized
scheme of the experiment is shown in Fig. 1
.
Spin-Echo NMR Imaging.
To assess glioma development in each rat and to select a slice
including the tumor for subsequent single voxel or SI experiments,
scout spin-echo images were obtained (slice thickness, 1 mm; TR, 3 s; TE, 80 ms; 128 x 128 pixels, 5 horizontal slices and
7 coronal slices).
Single Voxel Spectroscopy.
To follow changes in IEPA concentration and pHewithin the glioma, during the infusion of IEPA and the
subsequent 2 h from a volume in the center of the glioma,
experiments were made on three rats using point resolved spectroscopy
(Ref. 29
; TR, 3 s; TE, 40 ms; voxel size, 5 x 5 x 5 mm, 64 scans) taken at 5-min intervals.
At the end of the NMR measurements, the brains were excised and the
tumor dissected and frozen. The concentration of IEPA in an aqueous
extract of each tumor was measured by high-resolution
1H NMR spectroscopy.
1H SI.
The pulse sequence is shown in Fig. 2
. 1H spectroscopic images were acquired on four
rats using a spin-echo sequence with OVS (four slices), and CHESS
excitation (30)
for water saturation (3-lobe-sinc pulse, 8
ms duration, followed by crusher gradients of 2 ms, 20 mT/m). For IEPA
mapping (TE, 40 ms), a binomial pulse (31
, 32)
with a
frequency response centered on 8 ppm was used for excitation, and the
refocusing pulse was selective for a 2.5-mm horizontal slice
(5-lobe-sinc pulses, 6 kHz). For mapping other metabolites, both
excitation and refocusing pulses were slice selective (TE, 136 ms). WSI
were acquired at each TE for phase and frequency shift reference. The
field of view was 30 x 30 mm2.
Phase-encoding steps (32 x 32) were used with
acquisition weighting (Hanning window; Ref. 33
); TR was
3 s, so the total acquisition time was about 40 min. For WSI, TR
was 1 s, and the total acquisition time was 8 min. Bandwidth was
10 kHz with 1024-point acquisition in the time domain. Data were
filtered in the time domain using a Gaussian function resulting in a
10-Hz line broadening. Zero filling was performed in both spatial
dimensions to 64 points and in the spectral dimension to 2048 points
before three-dimensional Fast Fourier Transformation. After correction
for frequency shifts attributable to imperfect field homogeneity and
pre- and postperfusion data subtraction, the chemical shift of the IEPA
H2 resonance of each voxel spectrum, of which the signal:noise ratio
was >5, was obtained by standard nonlinear curve-fitting using the
complex Lorentzian line shape function. The chemical shift was
calculated on the assumption that the water peak was at 4.70 ppm; tests
showed that with this assumption the tCr peak consistently had a
chemical shift of 3.03 ppm in agreement with previous measurements
(34)
. pH was then obtained from the titration equation
(Eq. 1)
. In a small minority of voxels, two peaks within the range of
the H2 resonance were observed; in these cases, the larger peak was
fitted. In the spectra obtained with the longer TE, the areas of the
peaks assigned to lactate (1.31 ppm), NAA (2.03 ppm), tCho (3.23 ppm),
tCr (3.03 ppm), and lipids (1.26 ppm) were calculated. Peak areas in
each voxel were then normalized with respect to the water peak to
correct for instrumental variations between experiments.
|
Histology.
After the imaging experiment, the brain was excised and fixed in 10%
formol for 48 h and embedded in paraffin. Horizontal sections of
the glioma 3-µm thick were stained with H&E.
Errors of Measurement.
These have been considered in detail by van Sluis et al.(17)
. We estimate that the uncertainty in the
constants in the equation fitted to the in vitro pH
calibration points (Eq. 1
; Fig. 3e
) leads to an uncertainty of 0.035 pH units in our
estimates of the absolute scale of values of pHefor all of the in vivo measurements. In
addition, the uncertainty in the determination of the chemical shift of
the H2 peak in the spectrum for each voxel is estimated to lead to a
random error of 0.01 pH units for each voxel. Hence, our estimates of
absolute values of pHe, averaged over many
voxels, may be in error by 0.035 pH units. On the other hand, the
maximum error in the difference of pHe between
one voxel and another should be 2 x 0.01 = 0.02 pH units.
|
 | (2) |
(Xi - <X>)2 and
Sxy2 =
(Xi
-<X>)(Yi - <Y>), and Xi,
Yi are the values of the two
quantities in voxel i. |
RESULTS |
|---|
|
TopABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONREFERENCES |
|---|
is a T1-weighted image of a
horizontal section of a glioma-bearing rat brain acquired over 32 s starting 80 s after injection of a bolus of Gd-DOTA in the right
jugular vein. The marked hyperintensity in the periphery of the glioma
corresponded to well-perfused regions with leaky blood vessels
(35)
. Comparison with the histology confirmed that large
central parts of the glioma from which the signal was weak corresponded
to necrotic regions. Before the injection of Gd-DOTA, IEPA had been
infused during 60 min into the right jugular vein and
1H spectra with a 40-ms TE obtained from each
voxel of a slice (see Fig. 1
). After infusion of IEPA, the spectra from
the contralateral hemisphere were unchanged (Fig. 3a)
but
new peaks appeared in the glioma. A double peak near 7.2 ppm
corresponded to the H4 and H5 resonances of IEPA, which are relatively
insensitive to pH, and a peak near 8 ppm corresponded to the
pH-sensitive H2 resonance (arrow in Fig. 3b
; Refs. 16
, 17
). Analysis of the spectra showed that within the brain,
signals from IEPA reached our criterion amplitude (5 times the noise)
only within the glioma. The area under the H5 peak was measured for
each voxel and a map made (Fig. 3d)
. To relate the chemical
shift of the H2 peak to pH, we made an in vitro calibration
curve for IEPA in plasma at 37°C, which showed that the chemical
shift of the IEPA H2 proton varied from 7.72 ppm
(2) to 8.85 ppm
(1) with a
pKa of 6.566 ± 0.035 (Fig. 3e)
. By use of this curve, pHe was
calculated for each voxel and a map constructed (Fig. 3f)
.
This could only be done where IEPA was present, so there are no valid
measurements for normal tissue or for the center of the necrotic core.
The maps showed variations in pHe within the
glioma over a range of 0.4 pH units (Figs. 3f
and 6h)
, but no marked and significant general trend was
observed (e.g., from the center to the periphery). The mean
value of pHe measured in the regions infiltrated
by IEPA in each glioma was calculated; the mean value for the four
gliomas was 7.084 ± 0.017 (±SE). This is well below
the pH of the arterial blood, which was monitored in all of the
experiments and found to remain within the range 7.33–7.41 during the
IEPA imaging. In almost all of the voxels, only one H2 peak was
detected. In a few voxels there were two peaks, but both peaks always
corresponded to pHs more acidic than that of the circulating blood
(7.4). We failed to detect any association of the presence of two
peaks with any other feature of the gliomas. In agreement with the
majority of previous results in other tumors and tumor models made
using other methods of measurement, pHe in the C6
glioma was lower than the values most frequently reported for normal
brain tissue (7.3; Ref. 6
).
|
presents results that relate to the possibility that the presence of
IEPA in the extracellular clefts modified pHe.
pHe was measured in a single large voxel covering
about one-half the volume of the glioma. During the infusion of IEPA,
as the quantity in the glioma increased 3-fold,
pHe changed little and may have fallen slightly.
After the end of the perfusion (from 60 to 130 min), there was a small
but significant increase in pHe: a linear
regression gave pHe = 7.036 + 0.00127 x (t/min),
r2 = 0.87. Thus, in
comparing pHe in the presence of high and low
concentrations of IEPA, these results show that a high mean
concentration of IEPA was associated with a slightly more acidic
pHe.
Maps of Metabolites.
Fig. 5
shows typical 1H spectra obtained with a TE of
136 ms from normal tissue (a) and from the periphery
(b) and center (c) of a glioma. Because of the
J-coupling modulation, the lactate peak is negative and, for this
reason, its partial overlap with the lipid peak causes an apparent
shift of both chemical shifts and may distort the amplitudes when both
peaks are present. Because obtaining IEPA images was the main
methodological innovation of the work, the IEPA images were obtained
before the metabolic images, the latter being acquired 120–160 min
after the beginning of the IEPA infusion (Fig. 1)
. By this time, little
IEPA (20% of the maximum) still remained in the glioma (Fig. 4)
. If
residual IEPA contaminated the metabolic spectra, the peak most likely
to do so was the triple resonance from the IEPA methyl group. In
spectra obtained from extracts of gliomas (from other rats, not shown)
this IEPA peak was at 1.17 ppm. We conclude that it is unlikely that
the IEPA methyl peak contaminated the lactate peak in vivo.
However, the IEPA methyl group peak is closer to the methylene group
peak of the lipids (at 1.26 ppm). In the absence of certainty that the
lipid peak was free of contamination by the IEPA peak, we have not
considered the lipid distribution in this paper.
|
illustrates for a sample glioma the seven kinds of maps that were made.
As known for necrotic gliomas (35)
, the contrast agent
Gd-DOTA, or, more strictly, the water proton hypersignal it produces on
T1-weighted images, is predominantly confined to
the outer parts (Figs. 3c
and 6
, a and
b). The distribution of IEPA (Fig. 6c)
has some
similarity to that of Gd-DOTA. Comparison of the Gd-DOTA distribution
(Fig. 6, a and b)
and the histology with maps of
endogenous compounds detected by 1H SI with a TE
of 136 ms show that NAA was at a low concentration in gliomas (Fig. 6d)
. tCho or tCr tended to be present in the periphery (Fig. 6, e and f)
. Lactate was strongly present in
outer parts of the gliomas (Fig. 6g)
; its distribution in
the necrotic core could not be determined because of overlap of the
(negative) lactate peak and the (positive) lipid peak. Note that Fig. 6g
has not been corrected for this interference.
Voxel-by-Voxel Comparison of Distributions.
Because the resonance peaks of the 1H spectra
were normalized with respect to the water peak, the results for the
four gliomas were pooled. The various signals measured in the
rectangular areas including the gliomas (as in Fig. 6a
) were
then compared, voxel-by-voxel, for the various compounds and for
pHe. Sample scatter diagrams are shown in Fig. 7
. Fig. 7a
shows that tCho and tCr, both indicative of
metabolizing tissue, had, as expected, somewhat similar distributions.
Lactate posed a problem because of the overlap with the lipid peak.
Simulations of the two peaks showed that in voxels with no distinct
lipid peak, the maximum error in the area of the lactate peak was
<10%, and, for quantitative comparisons, we took into account only
these voxels with no detectable lipid. In those voxels where lactate
could be measured (i.e., where lipids were not detected), it
correlated with Gd-DOTA, a marker of extravasation through a
ruptured blood-brain barrier (Fig. 7b)
.
|
. The coefficients for NAA are not given, because coexistence of NAA and
a tumor marker occurred in only a small number of voxels at the
periphery of the gliomas, which probably included both tumoral and
normal tissue. Correlations for lactate were calculated only for voxels
where no lipid signal was apparent. The correlations in Table 1
will be
considered in the "Discussion."
|
|
DISCUSSION |
|---|
|
TopABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONREFERENCES |
|---|
0.05. During most of the time of the IEPA
imaging, IEPA was being cleared from the glioma (Figs. 1
and 4)
so the
concentration of IEPA in the plasma must have been less than that in
the nearby interstitium. Hence, although the blood volume fraction in
C6 gliomas may be higher than in 9L gliomas, the quantity of plasma
IEPA in a voxel was almost certainly <10% of the quantity of
interstitial IEPA. In the hypothetical case of no exchange of IEPA
molecules between the blood compartment and the interstitial
compartment within a voxel, one would observe, in the absence of noise,
two H2 peaks with two distinct chemical shifts corresponding to the two
pHs, 7.4 in the plasma and <7.1 in the interstitium. However, with
the signal:noise ratios of our spectra, the small peak corresponding to
the plasma compartment would have been excluded from the analysis, and,
indeed, no peak corresponding to pH 7.4 was detected. The other
extreme case, of exchange between the plasma and interstitial
compartments at a rate that is high compared with the frequency
difference corresponding to the two pH values is improbable but readily
analyzed. In this case, there would be a single peak at a frequency
corresponding to the weighted mean of the frequencies of the two
component peaks (37)
. With our worst case assumption that
the quantity of IEPA in the plasma is as great as 10% of the quantity
in the interstitium, then the single peak would be at the interstitium
frequency shifted 0.1 of the way toward the plasma frequency. Because
the calibration curve (Fig. 3e)
is nearly linear over the pH
range 7.1–7.4, this shift corresponds to 0.03 pH units. This is a very
conservative upper limit for the possible error introduced by the IEPA
in plasma.
Another possible source of systematic error is the possible
modification of pHe by the IEPA itself.
pHe is determined by the balance between the net
production (and subsequent extrusion) of proton equivalents by the
cells and the clearance of the extracellular proton equivalents to the
blood by diffusion through the extracellular clefts to the capillaries.
pH buffer can contribute to the latter process by facilitating
diffusion; flux of proton equivalents occurs not only by the diffusion
of H+ ions, which are present at a very low
concentration of 0.1 µM, but also by diffusion of the
buffer molecules, which may typically be present at concentrations
104
–105 times greater
(38)
. Hence, if the concentration of diffusible buffer
molecules in the extracellular space was increased (by addition of
IEPA), acid equivalents would be transferred more rapidly to the
capillaries and the pH in the interstitium would be increased to a
value closer to that of the blood (39)
. A
priori, this appears to be a problem. From the measurements of
[IEPA] in the glioma (Fig. 4)
we estimate the maximum
concentration of IEPA in the interstitium during the imaging to have
been 16 mM. The major mobile pH buffer is
HCO3-/CO2,
and the concentration of HCO3-
in equilibrium with a PCO2 of 35 mm Hg at pH 7.1
is 10.5 mM. Phosphates will contribute an
additional 4 mM. Although the diffusion
coefficient of IEPA is smaller than those of these endogenous anions
(Mr 211 versus 61
and 100), and it is not at its optimum pH for buffering
(pKa 6.5), IEPA may considerably increase
the concentration of diffusible buffers so that
pHe is shifted in the alkaline direction. This
predicted artifact might be reduced if there exist additional processes
undescribed in brain interstitium, such as the contribution to
facilitated diffusion by rotating proteins described in
vitro by Gros et al. (40)
. These arguments
predict that the greater the concentration of IEPA, the greater the
value of pHe. However, the experimental result of
Fig. 4
shows a change in the opposite direction: as the concentration
of IEPA in the glioma rose during infusion and then fell, not only did
pHe change very little, but
pHe was minimal, not maximal, when the IEPA
concentration was maximal (at the end of the 60 min infusion). Another
observation of the same type is that there was a significant negative
correlation between the spatial distribution of IEPA and
pHe (Table 1)
. In conclusion, we did not observe
the predicted alkalinizing effect of IEPA; either the effect was small,
or it was masked by other factors that remain to be identified.
Distributions of IEPA and Gd-DOTA.
Although both IEPA and Gd-DOTA were confined to the gliomas and had
their highest intensities in the peripheral regions, more IEPA than
Gd-DOTA was present in the center of the glioma, and the distribution
of IEPA correlated less than that of Gd-DOTA with the distributions of
tCho and tCr (Table 1)
. These observations are readily explained by the
much longer infusion time for IEPA, which would have allowed it to
diffuse farther away from the vasculature and into necrotic regions.
Distributions of tCho and tCr.
tCr, which is associated with storage of high energy phosphate
(41)
and tCho, which is associated with synthesis and
breakdown of membranes and with cell proliferation (42)
,
were strongly correlated (Table 1)
. As shown previously for human
tumors (43)
, tCho was highest in the periphery of the
gliomas (Fig. 6e)
, and it correlated strongly with Gd-DOTA,
indicating that it was in well-perfused regions (Table 1)
. Although the
correlation coefficients were smaller, tCr was also associated with
Gd-DOTA, and both tCho and tCr had positive correlations with
pHe. These results are coherent with the idea
that energy metabolism and proliferation are most active in
well-perfused, less acid parts of the glioma.
Distribution of Lactate.
In the voxels where the lactate signal was not contaminated by a lipid
signal, lactate correlated with tCho, tCr, and Gd-DOTA (Table 1)
,
i.e., it appeared to be present particularly in
well-perfused, actively metabolizing parts of the glioma, in agreement
with previous work (10
, 44)
. Production of lactate under
aerobic conditions is well-known for tumoral tissue (20)
and has been reported recently for several kinds of normal tissue,
including muscle (45)
, nerve (46)
, and brain
(19)
. In C6 gliomas it is known that lactate is rapidly
and heavily labeled from 14C glucose in blood
(21)
, and in C6 cells in culture lactate produced within
the cells tends not to enter the TCA cycle (47)
. Hence, it
is indeed to be expected that lactate is produced in the well-perfused
parts of the glioma even if they are adequately oxygenated. In the
center of the glioma, where we were unable to measure the relative
concentration of lactate, we would expect it to be at least as high as
the interstitial concentration in the outer parts, even if no lactate
were produced there because it would tend to be in diffusional
equilibrium. However, in the solid parts of the tumor, most of the
lactate is expected to be intracellular. This is not only because the
intracellular volume fraction is greater than the extracellular volume
fraction, but the intracellular concentration of lactate will be higher
than the extracellular concentration. Lactate can readily cross the
plasma membrane of most cells (including Ehlrich tumor cells; Ref.
48
) by passive cotransport with protons, so that the
concentrations of the neutral protonated form, Hlactate, inside and
outside the cell tend to equilibrate (22)
. Within each
compartment (intracellular and extracellular) the ratio of
[Hlactate]:[lactate-] is determined by the
pH according to the Henderson-Hasselbalch equation. It follows that
because [Hlactate]i 
[Hlactate]e, [total intracellular
lactate]/[total extracellular lactate]
[H+]e/[H+]i
(see Ref. 49
). pHi in C6 gliomas is
7.2 (50)
, so
[H+]e/[H+]i > 1, so [total lactate]i > [total lactate]e. Hence in the viable parts of
the glioma, total lactate will be mainly (80%) intracellular,
partly because intracellular space constitutes most of the volume, and
partly because [total lactate] is somewhat higher within this space.
Distribution of pHe in the Gliomas.
Although, as explained above, we cannot completely exclude the
possibility that the contribution of IEPA to facilitated diffusion of
protons through the extracellular clefts increases the absolute values
of pHe, spatial variations in
pHe should nevertheless be detected. On a scale
of mm, pHe showed variations on the order of 0.1
pH units (at least within that part of the glioma into which IEPA
diffused), in agreement with Ref. 10
. However, the
variations do not follow a regular geometry, and our data do not reveal
the presence of an average gradient of pHe from
the core toward the periphery in the C6 glioma model.
Although we found spatial variations of pHe of <0.1 pH unit mm-1, Helmlinger et al. (15) reported that measurements made on s.c. tumors with a fluorescent pH probe, which gave a higher spatial resolution, showed that gradients of pHe close to blood vessels can be very steep: >0.4 pH units/100 µm. Hence, the value we obtained for each voxel is the average of what may be a range of values of pHe.
Comparison of the Distributions of Lactate and pHe.
Although the formation of lactate from pyruvate consumes rather than
produces protons, the overall pathway of glycolysis from glucose to
lactate results in the net production of 2 ATP, which produce 1
H+ per ATP when dephosphorylated to ADP. One
glucose molecule also produces two lactate ions, which will be
cotransported out of the cell with two H+, so
although pHe will decrease there is no tendency
for pHi to change. In contrast to glycolysis,
full oxidation of glucose consumes protons that just balance the
production of H+ by hydrolysis of ATP; however,
there is production of CO2 (which reacts with
H2O to give H+) but at a
rate of only 1 CO2/6 ATP (51)
.
Because a cell requires a certain amount of ATP to function, it is
clear that if glycolysis replaces oxidative phosphorylation as a major
source of ATP, then more H+ equivalents will be
exported to the extracellular space. In general, this is true:
pHe is indeed more acid in lactate-producing
tumors than in normal tissue. Therefore, it is striking that within the
gliomas (at least in the parts with no detectable lipid signal) our
results show no significant negative correlation of lactate signal with
pHe (Table 1)
. A possible explanation is that
most lactate is produced in well-perfused regions with viable cells and
that in these regions H+ can diffuse away to the
blood stream more readily than lactate can.
We have demonstrated that a new probe molecule, IEPA, can be used for imaging pHe in a rat brain glioma model in vivo. This new technique made it possible to perform 1H SI in the same experiments. By comparing quantitatively the distributions of different metabolites, Gd-DOTA and pHe, no evidence was observed that pHe was significantly lower where lactate concentration was higher, and we reach the conclusion that lactate production is greatest in well-perfused parts of gliomas.
|
ACKNOWLEDGMENTS |
|---|
|
FOOTNOTES |
|---|
1 Supported in part by an Institut National de la
Santé et de la Recherche Médicale-Consejo Superior de
Investigaciones Cientificas Collaborative Grants and grants
08.1/0023/1997 and 08.1/0046/1998 (to S. C.); a Strategic Group
Grant (to P. B.) from the community of Madrid; and grants from La
Ligue contre le Cancer, l’Association pour la Recherche sur le Cancer,
and the Région Rhone-Alpes (to C. R.). 
2 Present address: Department of Radiology, Box
8131, Washington University School of Medicine, 660 South Euclid, St.
Louis, MO 63110.
3 To whom correspondence should be addressed, at
INSERM U438, CHU Pavillon B, BP 217, 38043 Grenoble Cedex 9, France.
4 The abbreviations used are: pHi,
intracellular pH; NMR, nuclear magnetic resonance; pHe,
extracellular pH; NAA, N-acetylaspartate; SI, spectroscopic imaging;
tCho, total choline-containing compounds; tCr, creatine and creatine
phosphate; CHESS, chemical-shift selective excitation; OVS, outer
volume saturation; TE, echo time; TR, repetition time; T2:
transverse relaxation time; pHa, arterial pH; Gd-DOTA,
gadolinium1,4,7,10-tetraazacyclododecane-N,N',N'',N'''-tetraaceticacid;
IEPA, (±)2-imidazole-1-yl-3-ethoxycarbonylpropionic acid; ppm, parts
per million; WSI, water spectroscopic imaging.
Received 2/27/01. Accepted 6/29/01.
|
REFERENCES |
|---|
|
TopABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONREFERENCES |
|---|
This article has been cited by other articles:
|
|
 |
|
  M.-F. Penet, A. P. Pathak, V. Raman, P. Ballesteros, D. Artemov, and Z. M. Bhujwalla Noninvasive Multiparametric Imaging of Metastasis-Permissive Microenvironments in a Human Prostate Cancer Xenograft Cancer Res., November 15, 2009; 69(22): 8822 - 8829. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
A. L. Vavere, G. B. Biddlecombe, W. M. Spees, J. R. Garbow, D. Wijesinghe, O. A. Andreev, D. M. Engelman, Y. K. Reshetnyak, and J. S. Lewis A Novel Technology for the Imaging of Acidic Prostate Tumors by Positron Emission Tomography Cancer Res., May 15, 2009; 69(10): 4510 - 4516. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 R. J. Gillies, I. Robey, and R. A. Gatenby Causes and Consequences of Increased Glucose Metabolism of Cancers J. Nucl. Med., June 1, 2008; 49(Suppl_2): 24S - 42S. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
P. Provent, M. Benito, B. Hiba, R. Farion, P. Lopez-Larrubia, P. Ballesteros, C. Remy, C. Segebarth, S. Cerdan, J. A. Coles, et al. Serial In vivo Spectroscopic Nuclear Magnetic Resonance Imaging of Lactate and Extracellular pH in Rat Gliomas Shows Redistribution of Protons Away from Sites of Glycolysis Cancer Res., August 15, 2007; 67(16): 7638 - 7645. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 G. G. Kovacs, A. Zsembery, S. J. Anderson, P. Komlosi, G. Y. Gillespie, P. D. Bell, D. J. Benos, and C. M. Fuller Changes in intracellular Ca2+ and pH in response to thapsigargin in human glioblastoma cells and normal astrocytes Am J Physiol Cell Physiol, August 1, 2005; 289(2): C361 - C371. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 X. Li, D. B. Vigneron, S. Cha, E. E. Graves, F. Crawford, S. M. Chang, and S. J. Nelson Relationship of MR-Derived Lactate, Mobile Lipids, and Relative Blood Volume for Gliomas in Vivo AJNR Am. J. Neuroradiol., April 1, 2005; 26(4): 760 - 769. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 N Raghunand, R A Gatenby, and R J Gillies Microenvironmental and cellular consequences of altered blood flow in tumours Br. J. Radiol., December 1, 2003; 76(suppl_1): S11 - S22. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
Y. Perez, H. Lahrech, M. E. Cabanas, R. Barnadas, M. Sabes, C. Remy, and C. Arus Measurement by Nuclear Magnetic Resonance Diffusion of the Dimensions of the Mobile Lipid Compartment in C6 Cells Cancer Res., October 15, 2002; 62(20): 5672 - 5677. [Abstract] [Full Text] [PDF]
|
|
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
| Cancer Research | Clinical Cancer Research |
| Cancer Epidemiology Biomarkers & Prevention | Molecular Cancer Therapeutics |
| Molecular Cancer Research | Cancer Prevention Research |
| Cancer Prevention Journals Portal | Cancer Reviews Online |
| Annual Meeting Education Book | Meeting Abstracts Online |
