Cocaine is pharmacologically active in the nonhuman primate fetal brain.

Abstract

Cocaine use during pregnancy is deleterious to the newborn child, in part via its disruption of placental blood flow. However, the extent to which cocaine can affect the function of the fetal primate brain is still an unresolved question. Here we used PET and MRI and show that in third-trimester pregnant nonhuman primates, cocaine at doses typically used by drug abusers significantly increased brain glucose metabolism to the same extent in the mother as in the fetus (approximately 100%). Inasmuch as brain glucose metabolism is a sensitive marker of brain function, the current findings provide evidence that cocaine use by a pregnant mother will also affect the function of the fetal brain. We are also unique in showing that cocaine's effects in brain glucose metabolism differed in pregnant (increased) and nonpregnant (decreased) animals, which suggests that the psychoactive effects of cocaine are influenced by the state of pregnancy. Our findings have clinical implications because they imply that the adverse effects of prenatal cocaine exposure to the newborn child include not only cocaine's deleterious effects to the placental circulation, but also cocaine's direct pharmacological effect to the developing fetal brain.

Full text

Page 1

Cocaine is pharmacologically active in the nonhuman
primate fetal brain
Helene Benvenistea,1, Joanna S. Fowlerb, William D. Rooneyc, Bruce A. Scharfd, W. Walter Backusa, Igor Izrailtyana,
Gitte M. Knudsene, Steen G. Hasselbalche, and Nora D. Volkowf
aDepartment of Anesthesiology, Stony Brook University, Stony Brook, NY 11794;bMedical Department, Brookhaven National Laboratory, Upton, NY 11973;
cAdvanced Imaging Research Center, Oregon Health and Science University, Portland, OR 97201;dOffice of Research, New Jersey Medical School, Newark, NJ
07101;eNeurobiology Research Unit, Rigshospitalet, Copenhagen University, 2100 Copenhagen, Denmark; andfNational Institute on Alcohol Abuse and
Alcoholism, Bethesda, MD 20892
Edited by Anissa Abi-Dargham, New York State Psychiatric Institute/Columbia University, New York, NY, and accepted by the Editorial Board December 11,
2009 (received for review August 22, 2009)
Cocaine use during pregnancy is deleterious to the newborn child,
in part via its disruption of placental blood flow. However, the
extent to which cocaine can affect the function of the fetal
primate brain is still an unresolved question. Here we used PET
and MRI and show that in third-trimester pregnant nonhuman
primates, cocaine at doses typically used by drug abusers signifi-
cantly increased brain glucose metabolism to the same extent in
the mother as in the fetus (∼100%). Inasmuch as brain glucose
metabolism is a sensitive marker of brain function, the current
findings provide evidence that cocaine use by a pregnant mother
will also affect the function of the fetal brain. We are also unique
in showing that cocaine’s effects in brain glucose metabolism dif-
fered in pregnant (increased) and nonpregnant (decreased) ani-
mals, which suggests that the psychoactive effects of cocaine are
influenced by the state of pregnancy. Our findings have clinical
implications because they imply that the adverse effects of prena-
tal cocaine exposure to the newborn child include not only
cocaine’s deleterious effects to the placental circulation, but also
cocaine’sdirectpharmacological
fetal brain.
effecttothedeveloping
brain function|in vivo|pregnancy|imaging|in utero
B
to occur only in the most mature third-trimester fetal brain.
Recently, discussions around the increase in fetal surgical pro-
cedures has raised the question of whether the fetus feels pain
and the need for adequate analgesia at least in third-trimester
fetuses when thalamocortical pain fibers are maturing (1). There
is also a serious lack of information in regards to whether drugs
taken by women during pregnancy, such as antidepressants or
drugs of abuse (e.g., cocaine and nicotine), are pharmacologi-
cally active in the fetal brain and linked with behavioral or
cognitive changes occurring in the child later in life. For exam-
ple, in the adult brain cocaine induces a feeling of euphoria, an
effect that is mediated in part by fast and supraphysiological
increases in dopamine in the brain’s reward circuits (including
the nucleus accumbens) secondary to cocaine’s blockade of the
dopamine transporter. However, the functional changes of
the fetus’ brain in response to maternal use of cocaine are
not known.
PETand2-deoxy-2-[18F]fluoro-D-glucose
measure brain glucose metabolism noninvasively, which serves as
a marker of brain function and has been used extensively to
evaluate the effects of psychoactive drugs (including drugs of
abuse) in the live brain (2–6). In the case of cocaine, PET studies
in cocaine abusers, showed that its i.v. administration produced a
14% global reduction in the cerebral metabolic rate of glucose
(CMRglu) that was associated with the subjective experience of
euphoria (4). Similar reductions in CMRglu have been demon-
strated in nonanesthetized nonhuman primates after short-term
cocaine self-administration (7) and in pregnant sheep, where
rain function and potentially cognitive experiences of the
fetus in utero are not well-understood and are likely (if at all)
(18FDG)can
cocaine was infused directly into the fetus using invasive
approaches (8). In the rodent, cocaine’s pharmacological action
in the fetus has also been demonstrated using quantitative
autoradiography (9, 10). However, the question of whether
cocaine can modify brain function in the primate fetal brain
when the mother is exposed to cocaine at a dose equivalent to
that abused by cocaine abusers remains unanswered.
Because the reinforcing effects of cocaine are associated with
its ability to block dopamine transporters, thereby increasing
dopamine (11), a prerequisite for cocaine to act as a drug rein-
forcer in the fetal brain would at least require the presence of
dopamine synthesis, functioning dopaminergic receptors, dop-
aminergic nerve fibers, and dopamine transporters. In vitro
studies on monkey and human tissues have demonstrated the
presence of dopamine receptors in second- and third-trimester
nonhuman primate fetuses (12), and dopamine transporters by
day 70 of gestation (13). There is also clear evidence of dop-
amine synthesis and spontaneous dopamine release in the fetal
rat brain, starting in the second half of uterine development (14).
In the nonhuman primate fetal brain, the information is more
limited; however, tyrosine hydroxylase-labeled axons and vari-
cosities have been demonstrated in the frontal cortex of rhesus
monkeys at the time of birth (15). Extrapolating from these data,
it appears that all of the necessary neurochemical components
are present in the fetal brain for cocaine to be pharmacologically
active and to have reinforcing effects (provided that sufficient
amounts of cocaine reaches the fetal brain).
We have developed a noninvasive method for visualizing
pharmacokinetic profiles of drugs concurrently in the mother
and fetus of nonhuman primates with PET and radio-labeled
ligands, including [11C]cocaine (16). With this imaging paradigm,
we have shown that there is significant uptake of cocaine in the
nonhuman primate fetal brain (≈80% of the cocaine uptake by
the mother’s brain). We now investigate the effect of maternal
intake of cocaine on the fetal brain function by noninvasively
measuring the fetal CMRglu, which serves as a marker of brain
function (17), using PET and18FDG in third-trimester pregnant
M. radiata. We hypothesized similar effects of cocaine on brain
glucose metabolism in the fetal and the maternal brain.
Author contributions: H.B., J.S.F., and N.D.V. designed research; H.B., W.D.R., B.A.S., W.W.
B., and I.I. performed research; H.B., J.S.F., I.I., and S.G.H. contributed new reagents/ana-
lytic tools; H.B., G.M.K., and S.G.H. analyzed data; and H.B., J.S.F., G.M.K., and N.D.V.
wrote the paper.
The authors declare no conflict of interest.
This article is a PNAS Direct Submission. A.A. is a guest editor invited by the Editorial
Board.
Data deposition: All PET and MRI imaging data sets are available via our database upon
request at: http://www3.bnl.gov/ifeinstein/viewPET.php.
1To whom correspondence should be addressed: Department of Anesthesiology, Stony
Brook University, Health Sciences Center, Level 4, Stony Brook, NY 11794. E-mail: Benve-
niste@bnl.gov.
1582–1587
| PNAS
| January 26, 2010
| vol. 107
| no. 4 www.pnas.org/cgi/doi/10.1073/pnas.0909585107

Page 2

Results
The main purpose of this study was to evaluate the effect of 1-
mg/kg cocaine, which is a dose in good correspondence to doses
used by cocaine abusers (18), administered i.v. to pregnant
nonhuman primates concurrently on the maternal and fetal
CMRglu. For this purpose, we applied a noninvasive dynamic
18FDG PET scan approach in which the
curve measured in the cardiac cavity of the mother is used as an
estimation of the arterial input function for calculation of
CMRglu in the maternal, as well as the fetal, brain. Twelve
healthy pregnant nonhuman primates (Macaca radiata) and five
healthy nonpregnant baboons were used in these studies and
divided into three groups (Table 1, and Methods). Group 1 (n =
8) comprised pregnant M. radiata primates who did not receive
cocaine (pregnant control group); group 2 (n = 4) comprised
pregnant M. radiata primates exposed to 1 mg/kg i.v. cocaine;
and group 3 comprised nonpregnant baboons tested twice with
and without exposure to 1 mg/kg i.v. cocaine and in whom the
CMRglu was calculated using a conventional invasive
arterial input function acquired during the dynamic PET scan
(Methods and Table 1).
18FDG time-activity
18FDG
Physiological Data. All subjects survived the scanning procedures
without morbidity or mortality and all of the pregnant M. radiata
delivered their babies without complications within 1 to 2 weeks
after the experiment.
There were no significant differences in the mean arterial
blood pressure during the dynamic PET scan between group 1
(control) subjects and group 2 subjects exposed to 1 mg/kg i.v.
cocaine. Importantly, the average i.v. anesthesia infusion rates
between the two groups were also within similar ranges (138 ± 32
μg/kg per min versus 150 ± 22 μg/kg per min) during the PET
scanning procedures, suggesting that all subjects were exposed to
the same anesthetic depth during the studies. Similarly, in group
3 (baboons) the mean arterial blood pressure and anesthesia
infusion rates were identical between the baseline control study
and during exposure to 1 mg/kg i.v. cocaine, and within the
infusion rate range of that used in the M. radiata. In the baboons,
the average plasma glucose during the first PET was 3.25 ± 0.23
mmol/L and was significantly higher (P = 0.028) during the
second PET scan acquired during exposure to 1 mg/kg cocaine
(3.49 ± 0.28 mmol/L).
Cerebral Metabolic Rate of Glucose Utilization. To verify that the
noninvasive approach for calculation of the CMRglu generated
reliable results, we first compared the CMRglu of the striatum
obtained from the pregnant M. radiata during control conditions
with the corresponding CMRglu calculated conventionally and
invasively via an
pregnant baboons. This analysis demonstrated that the CMRglu
18FDG arterial input function in the non-
values in the striatum measured in the pregnant M. radiata were
not statistically different from those in the baboons (10.1 ± 7.6
μmol/100 g per min vs. 11.6 ± 2.8 μmol/100 g per min, P > 0.05),
signifying that the noninvasive approach was assessing the
CMRglu correctly (Tables 2 and 3). Our data in the pregnant
nonhuman primates also showed that under control conditions
and with anesthesia, the CMRglu of the fetal brain was sig-
nificantly lower (40–50%) than that of the maternal brain (P <
0.04; Table 2).
It has been reported in the literature that an acute i.v. cocaine
challenge decreases the CMRglu in human subjects (4) and in
nonhuman primates (7). In accordance with prior studies, we
also showed that the same dose of cocaine decreased CMRglu in
the nonpregnant baboons (Fig. 1). Thus, in the nonpregnant
baboons the CMRglu was 11.6 ± 2.8 μmol/100 g per min under
control conditions, and decreased to 8.0 ± 1.9 μg/100 g per min
(P = 0.028) during the acute challenge with 1-mg/kg cocaine
(Table 3). In contrast to the CMRglu decreases observed in the
nonpregnant baboons, the acute i.v. challenge of cocaine sig-
nificantly increased the CMRglu in the pregnant maternal brain
(∼100%%) as well as in the fetal brain (∼100%) when compared
to the CMRglu in the pregnant controls (group 1) that did not
received cocaine (Fig. 2 and Table 2). Specifically, the maternal
CMRglu increased from 10.1 ± 7.6 μg/100 g per min to 20.5 ± 3.9
μg/100 g min (P = 0.037) during exposure to 1-mg/kg cocaine.
Similar to the maternal CMRglu, the CMRglu of the fetal brain
increased 2-fold (P < 0.01) from 5.1 ± 3.1 μg/100 g per min to
11.5 ± 2.5 μg/100 g min during exposure to 1 mg/kg cocaine
(Table 2).
Comparisons ofthe CMRglu in thepregnant M.radiata and the
nonpregnant baboons, both with and without cocaine exposure,
showed that even though they did not differ during control con-
ditions(nococaineexposure),theCMRgluwithcocaineexposure
was significantly higher in the M. radiata that in the nonpregnant
baboon (20.5 ± 3.9 vs. 8.0 ± 1.9 μg/100 g per min, P < 0.01).
Discussion
Our PET imaging data provide direct evidence that cocaine,
when administered i.v. to a pregnant nonhuman primate, is
pharmacodynamically active in the fetal brain. More specifically,
we are unique in showing that the nonhuman primate fetus’
brain function measured as changes in CMRglu increases sig-
nificantly in response to cocaine. Moreover, cocaine’s effects are
similar in the pregnant mother’s brain and in the fetal brain in
that the drug increases CMRglu to the same extent (≈100%), but
opposite to nonpregnant females in whom cocaine decreased
metabolism. We speculate that the opposite effects may reflect
the interaction between cocaine and high levels of circulating
hormones, including progesterone, estriol, and estetrol in the
pregnant female and fetus (19). Indeed, there is extensive data
Table 1. Experimental groups
Experimental group SpeciesPregnantPET18FDG Cocaine 1 mg/kg i.v.MRI scan
1 (n = 8)
2 (n = 4)
3 (n = 5)
M. radiata
M. radiata
Papio papio
Yes
Yes
No
Yes
Yes
Yes
No
Yes
Yes
Yes
Yes
No
Table 2. CMRglu of the fetal brain compared with that of the maternal brain
GroupMother CMRglu, μmol/min/100 g Fetus CMRglu, μmol/min/100 g
Group 1 (n = 8) (control)
Group 2 (n = 4) (1 mg/kg cocaine)
10.1 ± 7.6
20.5 ± 3.9†
5.1 ± 3.1*
11.5 ± 2.5*
*Fetal values that differ statistically from maternal values in group 1 (P < 0.05).
†Values of group 1 that differ statistically from group 2 (P < 0.05).
Benveniste et al. PNAS
| January 26, 2010
| vol. 107
| no. 4
| 1583
MEDICAL SCIENCES

Page 3

that both estrogen and progesterone modify the pharmacological
effects of cocaine (20, 21), and epidemiological data demon-
strating that cocaine use in humans addicted to cocaine is lowest
in the third trimester (22).
It is also possible that the increase in CMRglu we observed in
the maternal brain with cocaine is caused by a slower metabolism
of cocaine in pregnancy, resulting in higher than normal levels of
cocaine and its active metabolites when compared to the non-
pregnant female baboon brains. However, Jatlow showed no
differences in the pharmacokinetic profiles for cocaine and its
major metabolites (e.g., benzoylecgonine) in pregnant versus
nonpregnant subjects (23), eliminating this possibility. Further
work is needed to identify the mechanisms responsible for the
opposite effects of cocaine on CMRglu between pregnant and
nonpregnant animals.
In this study we chose to be as noninvasive as technically
possible for safety reasons, as this M. radiata colony is known to
periodically shed the Herpes B virus, which is structurally related
to herpes simplex viruses I and II of humans (24), and to avoid
manipulations that might interfere with normal delivery.
Therefore, instead of collecting maternal and umbilical arterial
blood for calculation of18FDG time activity curves (TAC), we
positioned the pregnant M. radiata transverse to the aperture of
the PET camera so that we could noninvasively obtain the
dynamic TACs from the maternal brain, maternal heart, and the
fetus simultaneously during the PET-scanning procedures. The
reliability of using the noninvasive approach for calculation of
the CMRglu was strongly supported by the finding that the
average CMRglu during control conditions of the pregnant M.
radiata was similar to the average CMRglu of the nonpregnant
female baboon brains calculated using an arterial input function.
We recognize that the use of nonpregnant female baboons as a
comparison control group to the pregnant M. radiata in our study
design comprises a potentially confounding factor, given the
possibility of interspecies differences. However, we were not able
to include or supplement with additional nonpregnant M. radiata
control subjects for technical as well as safety reasons.
The method of using a noninvasive-derived arterial input
function from the left ventricular blood pool to estimate the
metabolicrateofglucoseutilizationintheheartwith18FDGanda
Gjedde-Patlak graphical analysis was originally described by
Gambhir et al. (25). Ideally, the derived cardiac input function
should be corrected for spillover of activity between the left ven-
tricular blood pool and the myocardium using plasma samples
obtainedduringthelatterpartofthedynamicscan(26).Aswedid
not attempt to correct for partial volume effects (i.e., spillover
fromtheleftventricularchamberbloodpooltothemyocardium—
and vice versa—in the Macaca radiate), the cerebral input func-
tion, and consequently calculations of CMRglu, might therefore
be slightly underestimated by 10 to 20% (26). However, the
potential underestimation of the CMRglu would be the same for
group 1 (control) and group 2 (cocaine) and, therefore, the out-
come and conclusion of the analysis would still be valid.
TheCMRgluofthefetalbraincalculatedbasedonthematernal
leftventricleTACwasobservedtobeabout40%lowerthanthatof
the maternal brain. These findings are consistent with those
reportedintheliterature,whichalsoshowedlowerCMRgluinthe
fetal brain of guinea pigs (27) and rats (28) compared with the
maternal CMRglu. Even though one may have predicted higher
metabolism in the fetal than in the mother’s brain because the
brain is rapidly growing, this may be counteracted by lower func-
tional activity, alternative sources of energy (ketone bodies and
lactate), or greater sensitivity to the effects of anesthesia. Addi-
tionally, partial volume effects would tend to have a larger impact
in the smaller fetal brain but, if anything, this would lead to
underestimation of the fetal CMRglu.
All of the experiments in this study were performed in anes-
thetized animals and it is therefore possible that the pharmaco-
dynamic effect of cocaine was altered under these circumstances,
explaining the observed increase in CMRglu of the maternal as
well as the fetal brain. We used a combination of remifentanil and
propofol for anesthesia and both drugs are known to cross the
placenta: metabolites of remifentanil have been isolated from the
fetal circulation (29–31). Data reported in the literature does
confirm that the effect of cocaine on cerebral metabolism and
cerebral blood flow is critically dependent on the anesthetic used
(32–34) and, in some cases, different from that seen in the awake
state. On the other hand, our parallel data obtained in female
adult-nonpregnant baboons did demonstrate a cocaine-induced
decrease in CMRglu, as previously reported in awake human
subjectsandnonhuman primates,suggestingthat(i)thepropofol/
remifentanilanesthesiadidnotalterthepharmacodynamicprofile
ofcocaineasmeasuredbyCMRgluand(ii)thepharmacodynamic
effect of cocaine in third-trimester pregnancy is indeed different
when compared to the nonpregnant state.
Conclusions
The unique finding of this noninvasive imaging study is that an
acute challenge of cocaine at a dose (1 mg/kg i.v.) abused for
recreational purposes by cocaine abusers increased brain glucose
metabolism to the same extent in the maternal and the fetal
nonhuman primate brain. Inasmuch as brain glucose metabolism
Fig. 1.
during 1-mg/kg i.v. cocaine (Right) in an anesthetized nonpregnant female
baboon brain (transaxial views) from group 3. The scale bar at the bottom of
the PET images provides the color-coded hue correspondence of the glucose
utilization rate measured as micromole per 100 g per minute. It is clear from
the quantitative PET CMRglu images that the acute i.v. challenge of cocaine
reduces the CMRglu globally in the baboon brain. On average, in group 3, the
CMRglu at baseline was 11.6 ± 2 μg/100 g per min and decreased significantly
to 8.0 ± 1.9 μmol/100 g per min (P = 0.028) during 1 mg/kg cocaine exposure.
Parametric PET images of CMRglu acquired at baseline (Left) and
Table 3.
baboons
CMRglu and Kifor group 3 (n = 5) for nonpregnant
BaselineCocaine (1 mg/kg)
CMRglu (μmol/min/100 g)
Ki
11.6 ± 2.8
0.0189 ± 0.0052
8.0 ± 1.9*
0.0132 ± 0.0032*
Values are mean ± SD.
*Baseline values of group 3 that differ statistically from those obtained
during cocaine exposure (P < 0.05).
1584
| www.pnas.org/cgi/doi/10.1073/pnas.0909585107Benveniste et al.

Page 4

isasensitivemarkerofbrainfunction,thecurrentfindingsprovide
evidencethatcocaineusebya pregnantmotherwillalsoaffectthe
function of the fetal brain. We also show that cocaine’s effects on
thebrainglucosemetabolismdifferedinpregnantandnonpregnant
animals,whichsuggests thatthe psychoactive effectsofcocaine are
influenced by the state of pregnancy. Our findings have clinical
implications, as they suggest that the adverse effects of prenatal
cocaine exposure to the newborn child include not only cocaine’s
deleteriouseffectstoplacentalcirculation,butalsococaine’sdirect
pharmacological effect to the developing fetal brain.
Methods
Subjects. Twelve pregnant, female Bonnet macaques (M. radiata) from the
Primate Laboratory, Department of Psychiatry, State University of New York
(SUNY) Downstate and five nonpregnant adult female baboons (Papio
papio) from the vivarium at Brookhaven National Laboratory (BNL) were
used in this study (Table 1). The pregnant monkeys were judged to be in the
third-trimester gestational stage by the veterinarian after a brief physical
examination and ultrasound. The monkeys were transported to BNL 2 weeks
before the study and were housed according to National Institute of Health
guidelines at the BNL vivarium. The five adult female baboons (Papio papio)
were used as nonpregnant controls to validate our noninvasive approach to
quantify the cerebral metabolic rate of glucose utilization (CMRglu) using
18FDG PET (Table 1) and to compare the effects of acute cocaine on the
CMRglu between pregnant and nonpregnant animals. The experimental
protocol was approved by the Institutional Animal Care and Use Committees
at BNL and SUNY Downstate.
Anesthesia. All animals were anesthetized with a continuous infusion of
propofol (100–200 μg/kg per min; Diprivan, AstraZeneca) and Remifentanil
hydrochloride (0.1 μg/kg per min; ULTIVA). Mechanical ventilation was
implemented during the imaging procedures, as previously described (16,
35). Physiological parameters (blood pressure, heart rate, oxygen saturation,
body temperature, and end-tidal CO2) were monitored continuously and
kept within normal ranges. No muscle relaxant was used. At the conclusion
of imaging experiments, the anesthesia was discontinued and the animals
recovered spontaneous breathing within 2 to 5 min, were extubated, and
returned to their cages. All animals delivered their babies approximately
within 2 to 3 weeks of the experiment.
MRI of Pregnant M. radiata. All MRI data were acquired on a Varian/Siemens
4T instrument equipped with a high-performance gradient system (Sonata,
Siemens, Inc). A 27-cm inner diameter radiofrequency transceiver coil was
used for all studies. MRI data were collected under scan-synchronous ven-
tilation using a two-dimensional (2D) multislice dual spin-echo sequence;
proton density (TE15/TR3000) and T2-weighted (TE50/TR3000) MRIs were
typically acquired using 65 2-mm slices with a nominal in-plane resolution of
0.63 mm × 0.78 mm using a matrix of 512 × 256, and a total acquisition time
of 40 min. Only the maternal thorax, abdomen, and pelvis were included in
the field of view.
Dynamic FDG PET Studies. All PET scans were performed on an ECAT EXACT
HR+ tomograph (Siemens/CTI) with a 56-cm transaxial and 15.5-cm axial field
of view surrounded by detector blocks of 4 × 4-mm bismuth germanate
crystals 30 mm long.
The 12 pregnant M. radiata were divided into two groups: group 1 (n = 8)
served as controls; in group 2 (n = 4), the mother received an acute i.v.
challenge with 1mg/kg cocaine at the time of the18FDG injection. To obtain
the18FDG time course of activity in both maternal and fetal brain, as well as
the maternal heart, the animal was positioned transverse to the scanner axis
as previously described (16). Emission data were collected using acquisition
1 mg/kg Cocaine i.v.
Fetal Brain
Maternal Brain
Fetal Brain
Maternal Brain
Control
AB
Placenta
Placenta
Fig. 2.
brain time activity curve (as well as a fetal plasma glucose of 2.8 μmol/L) in a normal, pregnant M. radiata (group 1) and the corresponding T2-weighted MRI
image,demonstratingtheanatomicalpositionofthefetalbrain. Thescale barat thebottomofthePETimages provides thecolor-codedhuecorrespondence of
theglucoseutilizationratemeasuredasmicromoleper100gperminintheparametricimage(exceptfortheareacorrespondingtothematernalbladder,which
appearsasabrightredhomogenousarea).Thecolor-codedscalehelpsillustratethattheCMRgluofthefetalbrainapproximated7to8μmol/100gpermininthis
particularM.radiatafetus.(B)ParametricPETimageofthelocalmetabolicrateofglucose,alsocalculatedbasedonthefetalbraintimeactivitycurve(fetalplasma
glucose of 2.8 mmol/L) in a pregnant M. radiata exposed to 1-mg/kg cocaine i.v. (group 2); and the corresponding T2-weighted MRI image demonstrating the
position of the fetal brain. Again, the scale bar at the bottom of the PET images provides the color-coded hue correspondence of the glucose utilization rate
measuredasmicromoleper100gperminintheparametricPETimage(exceptforthematernalbladder)andisscaledidenticaltothecontrolimageinAfordirect
comparison.Thebluecolors representthelowvalues(0–10μmol/100 gpermin),andyellowandredcolors representhigher values(approximately>10–15μmol/
100gpermin).TheaverageCMRgluofthefetalbraininthisparticularM.radiatafetusapproximated10μmol/100gpermin.Notethetwosymmetricalhotspots
inthefetalbrain,whichappearinalocation likelytocorrespondtothestriatum.Onaverage,thefetalCMRgluincreased2-fold(P<0.01)from5.1±3.1μg/100g
perminto11.5±2.5μg/100gperminduringexposureto1-mg/kgcocaine,whilethematernalCMRgluincreasedfrom10.1±7.6μg/100gperminto20.5±3.9μg/
100 g per min (P = 0.037).
(A) A parametric PET image of an anesthetized pregnant M. radiata (sagittal view) of the local metabolic rate of glucose calculated based on the fetal
Benveniste et al. PNAS
| January 26, 2010
| vol. 107
| no. 4
| 1585
MEDICAL SCIENCES

Page 5

windows of 350 to 650 keV for energy and 12 ns for coincidence timing. To
maximize sensitivity, we used 3D mode (septa retracted), which gives a
spatial resolution at the center of the field of view of 4.6 mm transaxially
and 4.2 mm axially (using National Electrical Manufacturers Association NU-2
protocols). We obtained a separate 15-min68Ge transmission scan before
18FDG injection (1.5–5 mCi) to correct for attenuation by eliminating emis-
sion contamination, and in 2D mode (septa extended) to minimize accept-
ance of scattered events. PET data were processed in a fully quantitative
manner using the vendor ECAT software. Random coincidence events were
subtracted “on-line” using a delayed coincidence method. Attenuation cor-
rectionsinogramswere segmented before applicationtotheemissiondatato
reduce statistical noise propagation, and additional corrections for system
dead time, detector normalization, and scattered coincidences were per-
formed. The 3D emission sinograms were subsequently reduced to a set of 2D
sinograms using Fourier rebinning and then reconstructed into a 256 × 256
(transverse) × 63 (axial) image matrix using the attenuation-weighted
ordered-subsets expectation maximization algorithm. Six iterations with 16
subsetswereused,andtheimageswerepostfilteredwithaGaussiankernelof
4 mm in full width at half maximum to control noise. All studies were per-
formed using a dynamic acquisition sequence with the following18FDG time
framing:10×30s,5×60s,6×120s,1×180s,2×300s,1×480s,and1×720s.
The five nonpregnant female baboons (group 3) were exposed to a
baseline18FDG dynamic PET scan without cocaine and 2 to 4 weeks later a
repeat
coadministered i.v. at the same time as the18FDG. An arterial input function
via a femoral arterial line was obtained during both scans.
18FDG dynamic PET scan was performed with 1 mg/kg cocaine
Data Analysis. Pregnant M. radiata. All PET data analysis was performed using
PMOD (version 2.65; PMOD Technologies) and required several steps. First,
the whole-blood arterial input function was determined, which involved
accurately defining the lumen of the maternal left cardiac ventricle. This was
accomplished by visualizing the heart in a short-axis view on the earliest PET
time frames and using markers to carefully outline the left ventricle chamber
inthelong-axisview.Second,thestriatumwasidentifiedasdistinct“hot”spots
in the maternal forebrain after carefully viewing all three orthogonal views.
Third, for the fetus, the exact position and rotation of the fetal brain on the
anatomical MRI images were explored and used to guide identification of the
fetal forebrain and the hot spot areas within the striatum on the corre-
sponding PET images. For example, on the MRI images, all three orthogonal
planes were carefully investigated to identify the fetal brain in the sagittal
plane thatbest visualizedthe fetal forebrain, striatum, and cerebellum. Using
PMOD software, the forebrain/striatum and adjacent landmarks, such as the
placenta, was marked on the MRI image and identified on the corresponding
PET image. Next, a region of interest was positioned in the areas of hot spots
on the PET images assessed to correspond to the fetal striatum, with careful
motion adjustment if necessary. Finally, the maternal left heart-chamber TAC
was used as the input function for the subsequent kinetic analysis (Fig. 3). We
used the Gjedde-Patlak fitting routine in PMOD to determine the maternal and
fetalCMRgluandthecorrespondingslopeoftheregressioncurvefit.TheGjedde-
Patlak Plot estimations of CMRglu are critically dependent on correct values of
the plasma glucose concentration, because Kivaries inversely with the plasma
glucoseconcentration.Wedidnotmeasuretheplasmaglucoseconcentrationin
eitherthemotherorthefetusduringthePETstudy.However,frompreviousdata
itisknownthatthefastingplasma-glucoseconcentrationoftheadultM.radiata
isintherangeof2to4mmol/L.Weusedavalueof3.5mmol/Lforthemotherfor
both group 1and group 2. Furthermore,assuming that thefetalplasmaglucose
concentrationisknowntobeinaconstantratiotothemother’sbut∼20%lower
(36), we therefore used a plasma glucose concentration value of 2.8 mmol/L for
calculation of the fetal CMRglu. In the PMOD Gjedde-Patlak model (PKIN), the
originaltissueactivitymeasurementsaretransformedaccordingtothealgorithm
andaregressionlineisfittedtothecurvewithinarangethatcanbesetmanually,
so that only a certain portion of the curve is fitted. The ‘Lin’ start point was
adjusted to only fit the last three to four points of the curve (corresponding to
measurements taken >30 min following the18FDG injection).
Lumped constant. The lumped constant for 2-deoxyglucose (DG) in monkeys is
known [0.344 (37)]. However, as far as we are aware, the lumped constant for
18FDG in monkeys has not been measured. However, in rats the DG lumped
constant and FDG lumped constant are 0.48 (38) and 0.71 (39), respectively.
Therefore, for all our calculations, we used a LC of (0.71/0.48) × 0.344 = 0.507.
Furthermore, the LC is known to be relatively stable, also under conditions
with changes in blood ketone and lactate concentrations (40).
Baboons. The PET data analysis of the baboon brains was straightforward, as
both the arterial input function and plasma glucose was known. The analysis
was also performed using the PMOD Gjedde-Patlak model and by fitting to
the last three to four points of the transformed curves from the regions of
interest placed in the striatal area. We also used a lumped constant of 0.507
for these calculations.
Statistics.ThecalculatedparametersincludedtheCMRgluofthematernaland
fetal brains derived from the slope of the corresponding regression curve fits
(Ki) is shown in Fig. 3. All parameters are presented as mean ± SD. Differ-
ences between the CMRglu of the maternal and fetal brains were assessed
using a paired Student t test (two-tailed). Differences between pregnant
group 1 (control) and group 2 (cocaine) CMRglu values were assessed using
an unpaired Student t test (two-tailed). Differences between CMRglu values
in the nonpregnant baboon brains at baseline and after 1 mg/kg cocaine
was tested with a paired Student t test (one-tailed). The differences were
considered statistical significant at P < 0.05.
ACKNOWLEDGMENTS. We thank Donald Warner and David Alexoff for
assistance with the PET scan acquisitions; Pauline Carter, Payton King, and
Barbara Hubbard for help with the preparation, anesthesia, and imaging of
the nonhuman primates; and Dr. L. Rosenblum for allowing us access
to pregnant M. radiata from the Primate Laboratory at State University of
New York Downstate. This research was supported by Department of Energy
Office of Biological and Environmental Research (Brookhaven National Lab-
oratory Contract DE-AC02-98CH10886), National Institutes of Health, and
New York State Office of Science, Technology and Academic Research.
1. Lee SJ, Ralston HJ, Drey EA, Partridge JC, Rosen MA (2005) Fetal pain: a systematic
multidisciplinary review of the evidence. JAMA 294:947–954.
2. Bell RD, Alexander GM, Schwartzman RJ (1983) Methylphenidate decreases local
glucose metabolism in the motor cortex. Pharmacol Biochem Behav 18:1–5.
3. Breit S, Reimold M, Reischl G, Klockgether T, Wüllner U (2006) [(11)C]d-threo-
methylphenidate PET in patients with Parkinson’s disease and essential tremor. J
Neural Transm 113:187–193.
4. LondonED,etal.(1990)Cocaine-inducedreductionofglucoseutilizationinhumanbrain.
A study using positron emission tomography and [fluorine 18]-fluorodeoxyglucose.
Arch Gen Psychiatry 47:567–574.
5. Schreckenberger M, et al. (1999) “Ecstasy”-induced changes of cerebral glucose
metabolism and their correlation to acute psychopathology. An 18-FDG PET study. Eur
J Nucl Med 26:1572–1579.
6. VolkowND,etal.(1997)Effectsofmethylphenidateonregionalbrainglucosemetabolism
in humans: relationship to dopamine D2 receptors. Am J Psychiatry 154:50–55.
7. PorrinoLJ,LyonsD,SmithHR,DaunaisJB,NaderMA(2004)Cocaineself-administration
produces a progressive involvement of limbic, association, and sensorimotor striatal
domains. J Neurosci 24:3554–3562.
8. Burchfield DJ, Abrams RM (1993) Cocaine depresses cerebral glucose utilization in
fetal sheep. Brain Res Dev Brain Res 73:283–288.
Gjedde-Patlak Plot
•Plasma Glucose (mother) = 3.5 mmol/l
•Plasma Glucose (fetus) = 2.8 mmol/l
•Lumped constant = 0.507
Κi= 0.0179 (ml tissue/ml plasma)min-1
CMRglu= 12.0 μmol/min/100gm
Maternal brain
Time Activity Curves
Normalized plasma integral (min)
Fig. 3.
available software (PMOD) is illustrated. The maternal left-ventricle time-
activity curve is obtained from the maternal heart and used as the arterial
input function for calculation of CMRglu using the Gjedde-Patlak plot
approach. The graphical analysis of the Gjedde-Patlak plot is shown; the
regression line is fitted to the last three to four points of the curve and the
slopeofthelinearphaseoftheplotistheuptakerateconstantKi(mltissue/mL
plasma)min−1from which the CMRglu is calculated (41).
Calculation of the CMRglu in the maternal brain using commercially
1586
| www.pnas.org/cgi/doi/10.1073/pnas.0909585107Benveniste et al.