| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
|
|
|
|||||||
|
|||||||||
1 Uppsala University PET Centre, University Hospital Uppsala, S-75185 Uppsala; Departments of 2 Endocrinology and Physiology, 4 Clinical Neuroscience, and 6 Clinical Physiology, and 5 Karolinska Pharmacy, Karolinska Hospital, S-17176 Stockholm; and 3 Section Endocrinology, Department of Medicine, University Hospital of Trondheim, N-7006, Norway
 |
ABSTRACT |
|---|
 TOP
 ABSTRACT
 INTRODUCTION
MATERIALS AND METHODS
RESULTS
DISCUSSION
APPENDIX
REFERENCES
|
|---|
Using
R-
-[1-11C]hydroxybutyrate and positron emission
tomography, we studied the effect of acute hyperketonemia (range
0.7-1.7 µmol/ml) on cerebral ketone body utilization in six
nondiabetic subjects and six insulin-dependent diabetes mellitus (IDDM)
patients with average metabolic control (HbA1c = 8.1 ± 1.7%). An infusion of unlabeled R--hydroxybutyrate was
started 1 h before the bolus injection of
R--[1-11C]hydroxybutyrate. The time course of the
radioactivity in the brain was measured during 10 min. For both groups,
the utilization rate of ketone bodies was found to increase nearly
proportionally with the plasma concentration of ketone bodies (1.0 ± 0.3 µmol/ml for nondiabetic subjects and 1.3 ± 0.3 µmol/ml
for IDDM patients). No transport of ketone bodies from the brain could
be detected. This result, together with a recent study of the tissue
concentration of R--hydroxybutyrate in the brain by magnetic
resonance spectroscopy, indicate that, also at acute
hyperketonemia, the rate-limiting step for ketone body utilization
is the transport into the brain. No significant difference in transport
and utilization of ketone bodies could be detected between the
nondiabetic subjects and the IDDM patients.
-hydroxybutyrate; blood-brain barrier; positron emission
tomography
| |
INTRODUCTION |
|---|
|
TOP
ABSTRACT
INTRODUCTION
MATERIALS AND METHODS
RESULTS
DISCUSSION
APPENDIX
REFERENCES
|
|---|
KETONE BODIES SUPPLEMENT GLUCOSE as a fuel of the brain. The role of ketone bodies becomes important during special physiological conditions, such as long-term fasting. The resulting hyperketonemia is coupled to increased uptake and oxidation of ketone bodies in the brain. A parallel decrease in glucose oxidation maintains the total energy balance (8).
Type 1 diabetes in the poorly treated state is also characterized by hyperglycemia and hyperketonemia caused by insulinopenia. However, most type 1 diabetic patients also have 24-h levels of plasma ketone bodies during normal insulin treatment that are somewhat higher than those in nondiabetic subjects (7). Hyperketonemia will increase the uptake of ketone bodies in the brain, resulting in an increased oxidation of ketones. Such an increase would be additive to the excessive glucose uptake due to hyperglycemia and, unless counteracted by other regulation, would exacerbate the excess energy being delivered to the brain. Chronic hyperglycemia is known, at least in animals, to downregulate glucose transporters in the brain (13), thereby decreasing the glucose load to the brain. Analogously, ketone uptake and oxidation in the brain could also be subject to regulation in diabetes, although, to our knowledge, this has not previously been tested in subjects with type 1 diabetes.
A method has previously been developed for measuring regional cerebral
utilization of ketone bodies in humans with positron emission
tomography (PET) using R--[1-11C]hydroxybutyrate
(-[11C]HB) as tracer (1). The method was
applied in studies of healthy male subjects at normoketonemia. The
plasma concentration of R--hydroxybutyrate (-HB) was in the range
of 0.02-0.09 µmol/ml. Three main features of ketone body
utilization were observed. First, ketone body utilization was found to
increase almost linearly with increasing concentration of ketone bodies
in arterial plasma. Second, the uptake of ketone bodies could be well
described by a model with a single rate constant, indicating that the
uptake is essentially irreversible. Third, the tissue concentration of
ketone bodies was found to be very low, suggesting that the transport
across the blood-brain barrier (BBB) is the rate-limiting step for
ketone body utilization.
As early as 1971, Daniel et al. (4) reported that the transport of ketone bodies across the BBB in rat is essentially irreversible. Together with other studies on rats, this led to the hypothesis (2) that the carrier for ketone bodies is reversibly used for brain-blood transport of pyruvate and lactate.
The aim of the present study was to compare the ketone body utilization in nondiabetic subjects and subjects with insulin-dependent diabetes mellitus (IDDM) at hyperketonemia and to investigate whether the features of ketone body utilization previously observed at normoketonemia also persist at hyperketonemia.
| |
MATERIALS AND METHODS |
|---|
|
TOP
ABSTRACT
INTRODUCTION
MATERIALS AND METHODS
RESULTS
DISCUSSION
APPENDIX
REFERENCES
|
|---|
Subjects. The experimental procedure was approved by the Ethics and Radiation Safety Committees of the Karolinska Hospital and Institute. Before the subjects agreed to participate, they received written and oral information about the nature, purpose, and possible risks of the experiment. Six healthy male subjects served as a control group. Their mean age was 30.0 ± 7.4 yr (range 23-44 yr), and their body weight was 81 ± 10 kg. Six male patients with type 1 diabetes mellitus participated in the study. Their mean age was 32.0 ± 5.1 yr (range 23-38 yr), their body mass index was 24.5 ± 2.9 kg/m2, and their body weight was 78 ± 10 kg. The age at onset of the disease was 15.2 ± 10.3 yr, and the duration was 16.3 ± 6.9 yr. All of them were treated with multiple insulin injections except one patient, who was treated with continuous subcutaneous insulin infusion. Their usual daily dose of insulin was 37.0 ± 4.5 U of short-acting and 19.2 ± 4.8 U of long-acting insulin. The level of glycosylated hemoglobin (HbA1c) was 8.1 ± (SD) 1.7%. (The upper level of normal HbA1c was 5.6%.). None of the patients had recently experienced a serious hypoglycemic episode. Minimal signs of background retinopathy were present in four patients. One patient had macroalbuminuria but no other signs of nephropathy. Another patient was treated with antihypertensive drugs and a low dose of prednisolone because of glomerulonephritis due to systemic lupus erythematosus. Blood pressure was normal in all patients (<140/90 mmHg).
Experimental procedures.
To obtain ~1 mmol/l -HB in the plasma at steady state, a primed
infusion of a racemic mixture of unlabeled -HB was started 1 h
before the PET scan in all of the subjects. The proportion between the
R and S forms was close to 1:1. The priming dose, given during the
first 20 min (19), was twice the subsequent continuous
infusion dose, which lasted to the end of the PET scan. The priming
dose was 6 mg · kg
1 · min1
except in two cases (IDDM patients), when 3 and 4.5 mg · kg1 · min1,
respectively, were administered.
12 h after the last meal. The subjects with diabetes
mellitus reported to the ward at the Department of Endocrinology at 8 PM on the day preceding the study. Before dinner at around 7 PM, they
had injected their usual dose of short-acting insulin. Administration
of subcutaneous insulin was then discontinued, and an intravenous
infusion of insulin was started [50 U Actrapid Human (Novo Nordisk) in
250 ml of 0.9% saline solution]. This infusion was adjusted to
maintain blood glucose levels between 6 and 12 mmol/l during the night
and during the PET experiment. Before the PET measurements, and with
the subject under local anesthesia, a catheter was placed in the left
arteria brachialis for arterial blood sampling. A cannula was placed in
the right brachial vein for injection of the tracer.
-[1-11C]HB was synthesized in a two-step
stereo-specific synthesis starting with carrier-added
[11C]cyanide and R-propylene oxide. The total synthesis
time, including high-performance liquid chromatography purification,
was 45-50 min from the end of trapping. The radiochemical purity
of the products was >99% (20). The amount of
-[11C]HB administered was 100-400 MBq.
Immediately before the PET scan with -[11C]HB, a
transmission scan for determination of the attenuation correction was
performed. The PET scan after the bolus administration of
-[11C]HB was performed during a 10-min period, between
45 and 55 min after start of the infusion for one of the nondiabetic
subjects, and between 60 and 70 min for the remaining five healthy
volunteers and all of the six IDDM patients. The PET camera used was
the ECAT EXACT HR. The in-plane and axial resolutions are ~3.8 mm and
~4.0 mm (full width at half maximum), respectively (21). The measuring time was divided into 6 frames of 10 s, 3 frames of
20 s, 2 frames of 1 min, and 3 frames of 2 min; in all there were
14 frames. The camera was run in 2D mode in all experiments except
three. The results from these three runs in 3D mode did not differ from
the others. The radioactivity in arterial blood was sampled in 1-s
intervals by an automatic blood sampling system. To reduce the blood
loss, the sampler was used only during the first 5 min after the bolus
injection. The withdrawal rate was 5 ml/min. During this time, five
2-ml samples of arterial blood were drawn manually, and during the
remaining 5 min another four 2-ml samples were drawn manually. These
samples were measured in an NaI well counter that had been
cross-calibrated against the positron camera before the measurement. An
aliquot (0.5 ml) was taken from each sample and measured in the well
counter to determine the radioactivity concentration in whole blood.
The remaining part of each blood sample was centrifuged, and the
radioactivity in an aliquot of plasma (0.5 ml) was also measured in the
well counter. Samples were also taken at timed intervals to measure the
arterial concentrations of -HB, glucose, lactate, glycerol, and
insulin, and also to determine hematocrit, pH, and standard bicarbonate
concentration. The different concentrations were determined by standard
methods, as described by Grill et al. (6).
Data processing. The images of the radioactivity concentration in the form of matrices with 128 × 128 pixels for each of 47 slices, with a center-to-center distance of 3.125 mm, were reconstructed by standard software provided by the manufacturer.
To calculate the rate of uptake of-HB, the time course of
-[11C]HB in the arterial plasma, the "input
function" is required. The manual samples were taken from a three-way
connector at the catheter inserted in the artery. From the same
connector, blood was also pumped through the blood sampler. To be able
to compare the measurements made with the well counter and the blood
sampler, the transport time in the catheter from the connector to the
blood sampler has to be accounted for. The delay was 10 s with the withdrawal rate used (5 ml/min). The camera and blood data were synchronized by correcting for the difference in arrival times of the
radioactivity at the brain and at the blood sampler. The dispersion of
the input function in the catheter and in the peripheral artery was
corrected for with the aid of previous measurements.
On the basis of manual measurements, the ratio between the
radioactivity concentrations in the plasma and in the whole blood was
determined as a function of time. In accordance with the previous study
(1), the time dependence of this ratio could be well described by a straight line with a slope insignificantly different from zero. For the nondiabetic subjects, the intercept was found to be
1.21 ± 0.05 and the slope 0.00016 ± 0.00012 (average ± SD). For the IDDM patients, the intercept was found to be 1.19 ± 0.09 and the slope 
0.00004 ± 0.00033. With use of this
ratio, together with the two calibration factors, blood sampler against
well counter and well counter against camera, the time course of
-[11C]HB in arterial plasma was estimated second by
second from the whole blood measurements of the blood sampler. To
obtain the plasma concentration in the whole 10-min interval, the
manual measurements of -[11C]HB in the plasma
performed after the blood sampler measurements were added. The time
course of the arterial plasma concentration of
-[11C]HB was used as input function in the kinetic analysis.
Kinetic analysis.
Because of the low uptake of ketone bodies in the brain, determination
of regional ketone body utilization was not attempted in this study.
Identification of brain regions in PET studies using
-[11C]HB requires auxiliary measurement with magnetic
resonance imagery or some PET tracer with high uptake in the brain.
-[11C]HB is immediately
metabolized after entering the tissue and that the labeled metabolites
are irreversibly trapped during the measuring time (10 min). The model
contains two unknown parameters, the vascular fraction of cerebral
blood (CBV) and the parameter of interest, Kket,
the "accumulation rate constant" (see Eqs. A1 and A2).
Physiologically, there must be some concentration of ketone bodies in
the tissue (18). The second model applied, the "3k model," contains one reversible compartment for unmetabolized ketone
bodies and one irreversible compartment for metabolites. This model has
three rate constants (see Eq. A3). In the application of
this model, CBV was not fitted, but the value obtained with the 1k
model was used. Kket is obtained as
k1k3/(k2+k3).
Finally, we applied the Gjedde-Patlak analysis (5, 16). To
reduce the influence of residuals in the dominant blood peak in the
uptake, the data in the time interval 0-1 min were excluded from
the fit of the straight line. The parameters
Kket and DVapp were obtained as the
slope and y-intercept, respectively, of the fitted straight line. Also, in this case, the CBV value obtained from fit of the 1k
model was utilized. Thus the applied models contain one, two, and three
parameters in addition to CBV. For the 1k model and the Gjedde-Patlak
analysis, the parameters were obtained by linear regression, whereas
for the 3k model the parameters were estimated by using a nonlinear,
iterative, least squares method (12). These models were
also applied in the previous study at normoketonemia by use of the same
tracer (1).
In all models, the rate of ketone body utilization, CMRket,
is calculated as the product of Kket and the
concentration of -HB in the plasma, [-HB]plas. In
the previous report (1), this quantity was denoted
"primary CMRket." Pure -[11C]HB was
injected, but in blood this tracer rapidly becomes a mixture of
-[11C]HB and [11C]acetoacetate
([11C]AcAc). These compounds are transported across the
BBB with different rates, and therefore the measured CMRket
is a sum of the utilization rates of -HB and AcAc. With the method
used in this study, it is not possible to separate these two
components. In the previous study (1), the two utilization
rates were estimated with the aid of data from animal studies of ketone
body utilization in combination with certain assumptions (3,
10). It has been found that the plasma concentration of
[11C]AcAc in rats is very much lower than the
concentration of -[11C]HB (3, 10). If
this is also the case for humans, Kket reflects mainly utilization of -HB, and CMRket is close to the
rate of utilization of -HB, CMR
HB. This picture is
supported by a study by Hasselbalch et al. (9), in which
the plasma concentrations of unlabeled AcAc and -HB were measured in
humans. The ratio between the found values is 0.065 ± 0.074 at
normoketonemia and 0.111 ± 0.035 at hyperketonemia. In the
present study, we have compared the overall ketone body utilization of
nondiabetic subjects and IDDM patients with the aid of the primary
parameters Kket, CMRket, and
DVket, and we have not attempted to discriminate between the utilization of -HB and that of AcAc.
The ketone bodies enter the tricarboxylic acid cycle via acetyl-CoA,
and thereafter, tracer will be lost from the tissue mainly in the form
of [11C]CO2. In the previous study, it was
estimated that this loss leads to an underestimation of
CMRket by ~6% when a period of 10 min is used for
the parameter estimate. In the present study, we have not attempted to
correct for this loss. We assume that the effect of the loss is the
same for the nondiabetic subjects and the IDDM patients, and in the
comparisons of the results with data obtained in the previous study, we
also implicitly assume that the effect of the loss is the same at
normo- and hyperglycemia.
For the kinetic analysis, routines developed in-house to use the MATLAB
software were utilized.
| |
RESULTS |
|---|
|
TOP
ABSTRACT
INTRODUCTION
MATERIALS AND METHODS
RESULTS
DISCUSSION
APPENDIX
REFERENCES
|
|---|
Figure 1 shows the average
[-HB]plas as a function of time after start of the
infusion for the nondiabetic subjects and the IDDM patients. As can be
seen from Fig. 1, the start of the infusion (loading dose) caused an
immediate, rapid rise of [-HB]plas, but after 30 min
the rise was much less pronounced, and the concentration approached a constant level. As the error bars indicate, at a given
time point [-HB]plas varied considerably between the
experiments. In contrast, for each separate experiment, the time
variation of [-HB]plas was small during the PET
scan (60-70 min). In this interval, the difference between the
lowest and highest value of [-HB]plas within each
particular experiment was 3.3 ± (SD) 2.7% for the nondiabetic
subjects and 6.9 ± 2.5% and for the IDDM patients. In the same
time interval, [-HB]plas was found to be somewhat
higher for the IDDM patients than for the nondiabetic subjects
(1.28 ± 0.31 and 0.98 ± 0.33 µmol/ml, respectively). In
comparison, during the previous study, at normoketonemia
[-HB]plas was 0.04 ± 0.03 µmol/ml. Thus, in
the present study, [-HB]plas was, on the average, more
than 20 times higher than in the previous study at normoketonemia.
|
Figure 2, A-D,
shows examples of uptake curves and model fits for one control subject
(Fig. 2, A and C) and one IDDM patient (Fig. 2,
B and D). The results of applying the 1k model
and the Gjedde-Patlak analysis to the data are displayed. Figure 2
illustrates that both models give satisfactory fits of the data over
the whole time interval for both control subjects and IDDM patients.
For the healthy control subjects, Kket was found
to be 0.093 min1 with the 1k model, whereas for the IDDM
patients Kket was found to be 0.091 min1 with the same model. When the 3k model was applied
(not shown), the compartment of unmetabolized -[11C]HB
was always found to be small compared with the compartment of
metabolized -[11C]HB. Therefore, the fit is close to
the one obtained with the 1k model, and consequently the corresponding
values of Kket are also close to each other.
|
Figure 2, C and D, illustrates that the data in
the Gjedde-Patlak plot are well described by a straight line over the
whole measuring interval and that the intercept DVapp is
close to zero for all subjects. For many other tracers, it is possible
to distinguish an initial phase in the Gjedde-Patlak plot, before the
reversible part has reached equilibrium with the input function, but
clearly such a phase cannot be distinguished in this study. For the two distributions displayed in Fig. 2, C and D,
DVket was found to be 0.0011 and 0.0017, whereas
Kket was found to be 0.093 and 0.089, respectively. The initially large scatter in the data points is due to
difficulties in describing the dominant blood peak accurately in the
model. The radioactivity concentration is measured in arterial blood.
However, ~80% of the blood in the brain is venous, in which the
tracer has a different time course than in arterial blood.
The F-test and the Akaike information criterion (AIC) were applied to discriminate between the models. When testing the 1k model against the 3k model, i.e., when testing whether the two extra parameters are needed, significance was reached in 4 experiments (2 control subjects and 2 IDDM patients) out of 12 (level of significance 0.05) when the F-test was applied, and the 3k model was better in 7 experiments (4 controls and 3 IDDM patients) according to the AIC. It should be kept in mind that the F-test is strictly applicable (i.e., provides correct levels of significance) only for hypotheses that are linear in the parameters to be fitted. Clearly, the statistical analysis gives no preference for any of the models. Visually, good fits are obtained in all experiments for all models. For each experiment, nearly the same values of Kket were obtained with the different models. Unless otherwise stated, only Kket and the corresponding CMRket obtained with the 1k model are used in the following summary.
A summary of measured and fitted quantities (using the three models
discussed) is presented in Table 1. For
purposes of comparison, the corresponding quantities obtained in the
previous study at normoketonemia by use of the same tracer
(1) are also presented. Sample averages and standard
deviations are given. The values are averages over the brain, because
only average uptake curves have been used. It should be noted that the
presented values of concentrations in the plasma, such as [-HB],
in Table 1 are averages over the two or three measurements made during
the PET scan. In fact, most of the quantities stayed relatively
constant over the whole time interval (70 min) with the exceptions of
[-HB], [glucose], and [glycerol]. [Glucose] fell 3-10%
in the nondiabetic subjects and 15-35% in the IDDM patients in
the time period before the PET scan, but it then stayed constant within
a few percentage points during the PET scan. The concentration of
glycerol fell rapidly during the first 30 min, from 0.044 ± 0.019 (average ± SD) and 0.031 ± 0.009 µmol/ml for nondiabetic
subjects and IDDM patients, respectively. The averages reached in the
time interval of 60-70 min are presented in Table 1. Attempts to
fit the 3k model to the data often terminated with
k2 and/or k3 at the
allowed upper limit for these parameters (50 min1), and
therefore only upper bounds of DVket [=
k1/(k2+k3)]
can be given.
|
Figure 3 shows CMRket vs.
[-HB]plas for each individual in the two groups and
also the corresponding data obtained in the previous study of
nondiabetic subjects at normoketonemia. The displayed line is the
result of a linear regression analysis of all data in the present and
the previous studies. The slope, intercept, and R value were
found to be 7.9 ± 0.5, 0.39 ± 0.54, and 0.97, respectively.
With data only from the present study, the three parameters became
6.9 ± 1.3, 1.7 ± 1.6, and 0.85, respectively. In the
previous study, the same parameters were found to be 10.7 ± 0.8, 0.032 ± 0.040, and 0.99, respectively. Thus there is a weak
indication that the slope decreases with increasing
[-HB]plas, which means that the relationship between
[-HB]plas and CMRket is not perfectly
linear over the range of [-HB]plas in the present and
previous studies (0.02-1.74 µmol/ml). The regression
analysis of CMRket vs.
[-HB]plas reveals no significant difference between the two experimental groups of the present study. For the nondiabetic subjects, the slope 7.1 ± 2.1 and intercept 0.9 ± 2.2 are
obtained, whereas for the IDDM patients the slope 4.6 ± 1.9 and
intercept 5.1 ± 2.6 are obtained.
|
The relationship between plasma concentration and utilization is more
clearly seen in Fig. 4, which shows
Kket as a function of [-HB]plas
for all experiments in the present and the previous studies.
Kket and [-HB]plas
are measured independently of each other, and therefore, from a
statistical point of view, the data in Fig. 4 are easier to handle than
the data in Fig. 3, where the x- and y-variables
have a factor ([-HB]plas) in common. Clearly Kket has a tendency to decrease with increasing
[-HB]plas. Linear regression gives the slope
0.0025 ± 0.0006, intercept 0.0114 ± 0.0006, and R
value 0.76 for all experiments. The slope is significantly different from zero (P < 0.0005). For this regression,
all models gave consistent results. Figures 3 and 4 show that it is
difficult to distinguish any difference between the nondiabetic
subjects and the IDDM patients at hyperketonemia. When data from only
the present study are used, the slope 0.0023 ± 0.0021 and
intercept 0.011 ± 0.002 are obtained for the nondiabetic
subjects, whereas the slope 0.0033 ± 0.0017 and intercept
0.013 ± 0.002 are obtained for the IDDM patients. With the
present statistics, these regression lines are not significantly
different from each other or from the regression line obtained
from the combined data in the previous and present studies.
|
| |
DISCUSSION |
|---|
|
TOP
ABSTRACT
INTRODUCTION
MATERIALS AND METHODS
RESULTS
DISCUSSION
APPENDIX
REFERENCES
|
|---|
Tissue concentration of -HB
choice of model.
The 1k model gives a good fit of the uptake data for both nondiabetic
subjects and IDDM patients, indicating that the uptake of tracer across
the BBB is essentially irreversible. One parameter, Kket, describes the uptake in the tissue well
(Fig. 2, A and B). The successful fit of the
uptake data with this model implies that attempts to fit the data with
more complex models, such as the 3k model, result in unreliable
parameter estimates. This situation is a general feature of tracer
studies that use PET, because with this technique only the total
radioactivity concentration can be measured, which implies that the
compartmental structure that follows a nearly irreversible transfer is
very difficult to resolve. In particular, the PET experiment
alone gives unreliable information about DVket and the
related [-HB]tiss. In fact, as good fits as with the
1k and 3k models are obtained with a constrained 3k model that forces
DVket to be large.
-HB]tiss) can be interpreted in two ways.
1) With very low [-HB]tiss,
Jout also becomes low, even if the transport
capacity for ketone bodies out from the brain is appreciable, i.e.,
even if k2 is appreciable compared with
k3. 2) If this transport capacity is
drastically decreased, i.e., if k2 becomes very
low compared with k3,
Jout becomes low even if
[-HB]tiss is appreciable. Clearly PET experiments
using -[11C]HB cannot discriminate between the two
possibilities. Magnetic resonance (MR) spectroscopy provides
independent information about [-HB]tiss, which can be
utilized for this discrimination.
Recently, [-HB]tiss has been measured during acute
hyperketonemia in humans using high-field MR spectroscopy
(14). The -HB concentration was found to be 0.23 ± 0.10 µmol/ml when [-HB]plas was ~2 µmol/ml.
This measured -HB concentration can be explained by contributions
from blood and cerebrospinal fluid (CSF) spaces. Based on data obtained
after prolonged fasting (11), the summed contribution of
-HB in blood and CSF was estimated by Pan et al. (15)
to be 0.17 µmol/ml at the -HB plasma concentration of 3.5 µmol/ml, which is not significantly lower than the measured -HB
concentration in the tissue. This result indicates that the very low
Jout obtained in the present PET study at acute
hyperketonemia is an effect of a very low [-HB]tiss
and, consequently, indicates that the transport of ketone bodies into
the brain is the rate-limiting step. Nothing can be concluded
concerning the capacity for efflux (k2) of
ketone bodies across the BBB.
In contrast to the study at acute hyperketonemia, results obtained with
MR spectroscopy after 2 and 3 days of fasting (15) indicate that [-HB]tiss is considerable, close to 0.4 µmol/ml at [-HB]plas equal to 1 µmol/ml, and close
to 0.6 µmol/ml when [-HB]plas is close to 2 µmol/ml. Thus the utilization of ketone bodies in the brain is found
to be different at acute hyperketonemia and after fasting. However,
these results give no information about the efflux of ketone bodies
from the brain. Lactate and -HB share the same transport system
(monocarboxylic acid transporters) across the BBB, and during fasting,
the lactate concentration in the tissue increases from 0.69 ± 0.17 µmol/ml at normoketonemia to 1.31 ± 0.26 µmol/ml at
hyperketonemia (15). Therefore, it is possible that, due
to competition, the efflux of ketone bodies is suppressed in this
state. In contrast, during acute hyperketonemia, the lactate
concentration in the tissue was found to be 0.72 µmol/ml (15), close to the value found at normoketonemia. A PET
study with -[11C]HB as tracer would be suitable for
measurement of the efflux after fasting. If this process is
appreciable, the 1k model should not give a good fit, the 3k model
should give a DVket significantly larger than zero, and the
distribution of data in the Gjedde-Patlak plot should deviate from a
straight line during the initial time period.
It should be noted that the influence of AcAc has been neglected in the
analysis. From the MR experiment, DVHB is obtained, whereas with PET, DVket has contributions from both -HB
and AcAc. Because [AcAc]plas is low compared with
[-HB]plas (2, 9, 10), it is likely that
[AcAc]tiss is lower than [-HB]tiss. Furthermore, we have not corrected for loss of
[11C]CO2 from the tissue. Such a correction
cannot change the conclusions made, however, because it should give
somewhat steeper slopes in the Gjedde-Patlak analysis, which in turn
should tend to give even lower values of DVapp.
A low [-HB]tiss value implies that the pool of
unmetabolized -HB in the tissue will rapidly reach equilibrium with
-HB in the plasma. Therefore, it is expected that the degree of
steady state reached for [-HB]plas, shown in Fig. 1,
reflects the degree of steady state reached for
[-HB]tiss.
Metabolic rate of -HBcomparison between control subjects and
IDDM patients.
The data in Table 1 show that the values of the rate constant for net
utilization Kket, estimated with different
models, are close to each other. The good fits using the 1k model and the Gjedde-Patlak analysis show that, also at acute hyperketonemia, the
rate of utilization is very close to the unidirectional influx of
-HB across the BBB. Therefore, in this case, influx, uptake, and
utilization are synonymous.
-HB]plas considered,
CMRket is found to be nearly but not completely
proportional to [-HB]plas for both nondiabetic
subjects and for IDDM patients. Compared with the previous study at
normoketonemia, [-HB]plas is 23 times larger in this
study for the nondiabetic subjects and 30 times larger for the IDDM
patients, whereas the corresponding CMRket values are only
16 and 22 times larger, respectively. CMRket for the IDDM
patients is, on the average, 39% larger than CMRket for
the nondiabetic subjects, but this difference can be explained by the
fact that [-HB]plas is, on the average, 31% larger
for the IDDM patients than for the nondiabetic subjects. The regression analysis shows that the Kket values in the two
groups are not significantly different from each other (Fig. 3). It
should be noted that the S-isomer, included in the racemic mixture
(50%) and used in the present study, may be metabolized somewhat
less and might have a different fate compared with the R-isomer
(17).
The results indicate that the net utilization of ketone bodies is far
from saturation in both the nondiabetic subjects and the IDDM patients.
The origin of the small deviation from strict proportionality between
[-HB]plas and CMRket (decreasing
Kket) might be limitations in the transport
capacity across the BBB. Another reason might be an increasing rate of
loss of [11C]CO2 from the tissue with
increasing CMRket. According to Hasselbalch et al.
(9), cerebral blood flow (CBF) increases as an effect of
acute hyperketonemia. CBF was found to increase from 0.51 ± 0.09 to 0.71 ± 0.17 ml · g1 · min1 at
[-HB]plas equal to 0.31 ± 0.17 and 2.16 ± 0.42 µmol/ml, respectively. Increasing CBF implies more effective
removal of [11C]CO2 produced in the tissue,
counteracting the increased rate of accumulation of tracer in the
tissue due to increased metabolism and
[11C]CO2 production. It is difficult to judge
the net effect, but it cannot be excluded that hyperketonemia causes a
larger proportion of the tracer to be lost from the tissue, which would
explain the small decrease of Kket with
increasing [-HB]plas.
Hasselbalch et al. (9) studied cerebral uptake of ketone
bodies before and during the infusion of -HB in healthy volunteers. -HB was infused during 140 min before the study.
CMRHB was measured by the Kety-Schmidt technique and
found to be 11.1 ± 12.1 nmol · ml1 · min1 at
[-HB]plas equal to 0.31 ± 0.17 µmol/ml,
and 56.0 ± 22.5 nmol · ml1 · min1 at
[-HB]plas equal to 2.16 ± 0.42 µmol/ml. The
corresponding values for uptake of AcAc were 0.00 ± 0.01 nmol · ml1 · min1 and
24.9 ± 4.17 nmol · ml1 · min1,
respectively. These data are not directly comparable with our results;
however, in agreement with the tendency observed in our study, the
increase in CMRHB (a factor of 5) was found to be
somewhat smaller than the increase in [-HB]plas (a
factor of 7).
A major objective of the present study was to test whether chronic
hyperglycemia and the tendency for hyperketonemia, typical for type 1 diabetes, would affect ketone body metabolism in the brain. This study
does not provide evidence to support such a metabolic effect. The
patients included in the study had average metabolic control, as
assessed by their levels of HbA, and the results are therefore
applicable to a majority of type 1 diabetic patients. To detect
differences relating to chronic ketonemia, we would have had to recruit
type 1 diabetic patients with markedly poor control. Such a study,
however, was precluded by ethical concerns and by a scarcity of
patients in the poor control category who were both willing and able to
participate. Hence, we cannot exclude that worse metabolic control,
i.e., more severe hyperglycemia and a stronger tendency for
hyperketonemia than in the presently studied patients, might cause
abnormalities in the uptake and metabolism of ketone bodies in the brain.
Conclusions.
With plasma concentration of -HB in the range 0.02-1.74
µmol/ml, the brain tissue was found to react to increased
availability of ketone bodies by an increased net utilization of these
compounds. There was no sign of saturation of this process. This holds
true for both nondiabetic subjects and type 1 diabetic patients with average metabolic control. The transport of ketone bodies across the
BBB is found to be essentially irreversible. Together with a recent
study with MR spectroscopy showing low tissue concentration of -HB
(15), the findings of this study indicate that, also at
acute hyperketonemia, the transfer from blood to brain is the rate-limiting step in ketone body utilization.
| |
APPENDIX |
|---|
|
TOP
ABSTRACT
INTRODUCTION
MATERIALS AND METHODS
RESULTS
DISCUSSION
APPENDIX
REFERENCES
|
|---|
Kinetic Models Used
In the 1k model, it is assumed that the transfer of-[11C]HB is irreversible and that the tracer is
trapped in the tissue during the measuring time (10 min). The model
contains a single parameter, Kket, the
accumulation rate constant, and the time course of the tracer in the
tissue is governed by the simple expression
|
(A1) |
|
(A2) |
In the 3k model, with one reversible and one irrreversible tissue
compartment and with three rate constants,
Ctiss(T) is expressed as
![C<SUB>tiss</SUB>(<IT>T</IT>)<IT>=k</IT><SUB>1</SUB><IT> · k</IT><SUB>2</SUB><IT>/</IT>(<IT>k</IT><SUB>2</SUB><IT>+k</IT><SUB>3</SUB>)<IT> · </IT><LIM><OP>∫</OP><LL>0</LL><UL><IT>T</IT></UL></LIM><IT>e</IT><SUP>[(<IT>k</IT><SUB>2</SUB>+<IT>k</IT><SUB>3</SUB>)<IT> · </IT>(<IT>x</IT>−<IT>T</IT>)]</SUP><IT> · </IT>C<SUB>pla</SUB>(<IT>x</IT>)d<IT>x</IT>](/content/vol283/issue1/fulltext/E20/img003.gif)
|
(A3) |
|
-HB] · DVket. The accumulation rate
constant Kket,
k1 · k3/(k2 + k3), measures the efficiency of the metabolism
of -HB, and the corresponding CMRket is
[-HB]plas · Kket. The
unidirectional rate of influx across the BBB,
Jin, is
[-HB]plas · k1.
In the Gjedde-Patlak analysis, the operational equation is
|
(A4) |
|
| |
ACKNOWLEDGEMENTS |
|---|
This study was supported by grants from the Swedish Medical Research Council, MFR, project nos. K98-04F-12394-01, K97-04P-11319-03A, 04540, and 8276, and the Swedish Diabetes Association.
| |
FOOTNOTES |
|---|
Address for reprint requests and other correspondence: G. Blomqvist, Uppsala Univ. PET Centre, UAS, 75185 Uppsala, Sweden (E-mail: gunnar.blomqvist{at}pet.uu.se).
The costs of publication of this article were defrayed in part by the payment of page charges. The article must therefore be hereby marked "advertisement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
First published February 26, 2002;10.1152/ajpendo.00294.2001
Received 5 July 2001; accepted in final form 22 February 2002.
| |
REFERENCES |
|---|
|
TOP
ABSTRACT
INTRODUCTION
MATERIALS AND METHODS
RESULTS
DISCUSSION
APPENDIX
REFERENCES
|
|---|
1.
Blomqvist, G,
Thorell JO,
Ingvar M,
Grill V,
Widén L,
and
Stone-Elander S.
Use of R--[1-11C]hydroxybutyrate in PET studies of regional cerebral uptake of ketone bodies in humans.
Am J Physiol Endocrinol Metab
269:
E948-E959,
1995
2.
Cremer, JE,
Braun LD,
and
Oldendorff WH.
Changes during development in transport processes of the blood-brain barrier.
Biochim Biophys Acta
448:
633-637,
1976[Medline].
3.
Cremer, JE,
and
Heath DF.
The estimation of rates of utilization of glucose and ketone bodies in the brain of the suckling rats using compartmental analysis of isotopic data.
Biochem J
142:
527-544,
1974[Medline].
4.
Daniel, PM,
Love ER,
Moorhouse SR,
Pratt OE,
and
Wilson P.
The movement of ketone bodies, glucose, pyruvate and lactate between the blood and the brain of rats.
J Physiol
274:
22P-23P,
1971.
5.
Gjedde, A.
High- and low-affinity transport of D-glucose from blood to brain.
J Neurochem
36:
463-471,
1981.
6.
Grill, V,
Gutniak M,
Björkman O,
Lindqvist M,
Stone-Elander S,
Seitz RJ,
Blomqvist G,
Reichard P,
and
Widén L.
Cerebral blood flow and substrate utilization in insulin-treated diabetic subjects.
Am J Physiol Endocrinol Metab
258:
E813-E820,
1990
7.
Harano, Y,
Kosugi K,
Hyosu T,
Suzuki M,
Hidaka H,
Kashowagi A,
Uno S,
and
Shigeta Y.
Ketone bodies as markers for type 1 (insulin dependent) diabetes and their value in the monitoring of diabetic control.
Diabetologia
26:
343-348,
1984[Medline].
8.
Hasselbalch, SG,
Knudsen GM,
Jakobsen J,
Pinborg Hageman L,
Holm S,
and
Paulson OB.
Brain metabolism during short-term starvation in humans.
J Cereb Blood Flow Metab
14:
125-131,
1994[Medline].
9.
Hasselbalch, SG,
Madsen PL,
Hageman LP,
Olsen KS,
Justesen N,
Holm S,
and
Paulson OB.
Changes in cerebral blood flow and carbohydrate metabolism during acute hyperketonemia.
Am J Physiol Endocrinol Metab
270:
E746-E751,
1996
10.
Hawkins, RA,
and
Mans AM.
Regional blood-brain barrier transport of ketone bodies in portacaval-shunted rats.
Am J Physiol Endocrinol Metab
261:
E647-E652,
1991
11.
Lamers, KJB,
Doesburg WH,
Gabreels FJM,
Romsom AC,
Lemmens WAJG,
Wevers RA,
and
Renier WO.
CSF concentration and CSF/blood ratio of fuel related components in children after prolonged fasting.
Clin Chim Acta
167:
135-145,
1987[Medline].
12.
Marquardt, DW.
An algorithm for least-squares estimation on nonlinear parameters.
J Soc Indust Appl Math
11:
431-441,
1963.
13.
McCall, AL,
Fixman LB,
Fleming N,
Tornheim K,
Chick W,
and
Ruderman NB.
Chronic hypoglycemia increases brain glucose transport.
Am J Physiol Endocrinol Metab
251:
E442-E447,
1986
14.
Pan, JW,
Lee JH,
Telang FW,
Gabriely I,
de Graaf RA,
Sammi MK,
Stein DT,
Rothman DL,
and
Hetherington HP.
Measurement of human brain -hydroxybutyrate in acute hyperketonemia.
Proc Int Soc MR Med
9:
384,
2001.
15.
Pan, JW,
Rothman DL,
Behar KL,
Stein DT,
and
Hetherington HP.
Human brain -hydroxybutyrate and lactate increase in fasting-induced ketosis.
J Cereb Blood Flow Metab
20:
1502-1507,
2000[ISI][Medline].
16.
Patlak, C,
Blasberg RG,
and
Fenstermacher JD.
Graphical evaluation of blood-to-brain transfer constants from multiple time uptake data.
J Cereb Blood Flow Metab
3:
1-7,
1983[ISI][Medline].
17.
Robinson, A,
and
Williamson D.
Physiological role of ketone bodies as substrates and signals in mammalian tissues.
Physiol Rev
60:
143-187,
1980
18.
Ruderman, NB,
Ross PS,
Berger M,
and
Goodman MN.
Regulation of glucose and ketone body metabolism in brain of anaesthetized rats.
Biochem J
138:
1-10,
1974[ISI][Medline].
19.
Scherwin, RS,
Hendler RG,
and
Felig P.
Effect of ketone body infusions on amino acid and nitrogen metabolism in man.
J Clin Invest
52:
1382-1390,
1975.
20.
Thorell, JO,
Stone-Elander S,
König WA,
Halldin C,
and
Widén L.
Synthesis of R- and S[1-11C]--hydroxybutyric acid.
J Labelled CompD Radiopharm
6:
709-718,
1991.
21.
Wienhard, K,
Dahlbom M,
Eriksson L,
Michel C,
Bruckbauer T,
Pietrzyk U,
and
Heiss WD.
The ECAT EXACT HR: performance of a new high resolution positron scanner.
J Comput Assist Tomogr
18:
110-118,
1994[ISI][Medline].
This article has been cited by other articles:
|
|
 |
|
  S. L. Wootton-Gorges, M. H. Buonocore, N. Kuppermann, J. Marcin, J. DiCarlo, E. K. Neely, P. D. Barnes, and N. Glaser Detection of Cerebral {beta}-Hydroxy Butyrate, Acetoacetate, and Lactate on Proton MR Spectroscopy in Children with Diabetic Ketoacidosis AJNR Am. J. Neuroradiol., May 1, 2005; 26(5): 1286 - 1291. [Abstract] [Full Text] [PDF]
|
|
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
| Visit Other APS Journals Online |
) have been placed in the middle
of the measuring intervals. The 1k model (A and
B) has 2 components, the vascular component and the tissue
compartment. In the Gjedde-Patlak analysis (C and
D), contribution from the vascular blood component
CBV · Cblo(t) has been subtracted before
the y-variable,
Ctiss(t)/Cpla(t), is
calculated and plotted against the x-variable
Cpla(x)dx/Cpla(t),
"normalized time." Regression lines obtained by fitting the 2 distributions in the interval 1-10 min are displayed.
) and IDDM patients
(
) at hyperglycemia and in the group of nondiabetic
subjects at normoketonemia (