Effects of Cell Phone Radiofrequency Signal Exposure on Brain Glucose Metabolism

PRELIMINARY
COMMUNICATION
Effects of Cell Phone Radiofrequency Signal
Exposure on Brain Glucose Metabolism
Nora D. Volkow, MD
Context The dramatic increase in use of cellular telephones has generated concern
Dardo Tomasi, PhD
about possible negative effects of radiofrequency signals delivered to the brain. How-
Gene-Jack Wang, MD
ever, whether acute cell phone exposure affects the human brain is unclear.
Paul Vaska, PhD
Objective To evaluate if acute cell phone exposure affects brain glucose metabo-
lism, a marker of brain activity.
Joanna S. Fowler, PhD
Design, Setting, and Participants Randomized crossover study conducted be-
Frank Telang, MD
tween January 1 and December 31, 2009, at a single US laboratory among 47 healthy
Dave Alexoff, BSE
participants recruited from the community. Cell phones were placed on the left and right
ears and positron emission tomography with (18F)fluorodeoxyglucose injection was used
Jean Logan, PhD
to measure brain glucose metabolism twice, once with the right cell phone activated (sound
Christopher Wong, MS
muted) for 50 minutes (“on” condition) and once with both cell phones deactivated (“off”
T
condition). Statistical parametric mapping was used to compare metabolism between on
HE DRAMATIC WORLDWIDE IN-
and off conditions using paired t tests, and Pearson linear correlations were used to verify
crease in use of cellular tele-
the association of metabolism and estimated amplitude of radiofrequency-modulated elec-
phones has prompted con-
tromagnetic waves emitted by the cell phone. Clusters with at least 1000 voxels (volume
cerns regarding potential
8 cm3) and P
.05 (corrected for multiple comparisons) were considered significant.
harmful effects of exposure to radiofre-
Main Outcome Measure Brain glucose metabolism computed as absolute me-
quency-modulated electromagnetic fields
tabolism (µmol/100 g per minute) and as normalized metabolism (region/whole brain).
(RF-EMFs). Of particular concern has
Results Whole-brain metabolism did not differ between on and off conditions. In con-
been the potential carcinogenic effects
trast, metabolism in the region closest to the antenna (orbitofrontal cortex and temporal
from the RF-EMF emissions of cell
pole) was significantly higher for on than off conditions (35.7 vs 33.3 µmol/100 g per
phones. However, epidemiologic stud-
minute; mean difference, 2.4 [95% confidence interval, 0.67-4.2]; P=.004). The increases
ies of the association between cell phone
were significantly correlated with the estimated electromagnetic field amplitudes both for
use and prevalence of brain tumors have
absolute metabolism (R=0.95, P .001) and normalized metabolism (R=0.89; P .001).
been inconsistent (some, but not all,
Conclusions In healthy participants and compared with no exposure, 50-minute cell
studies showed increased risk), and the
phone exposure was associated with increased brain glucose metabolism in the re-
issue remains unresolved.1
gion closest to the antenna. This finding is of unknown clinical significance.
RF-EMFs emitted by cell phones are
JAMA. 2011;305(8):808-814
www.jama.com
absorbed in the brain2 within a range that
phones have yielded variable results.6 For
The objective of this study was to as-
could influence neuronal activity.3 Al-
example, imaging studies that used
sess if acute cell phone exposure af-
though the intensity of RF-EMFs is very
positron emission tomography (PET) to
fected regional activity in the human
low, the oscillatory frequencies corre-
measure changes in cerebral blood flow
brain. For this purpose we evaluated the
spond to some of the oscillation frequen-
(CBF) with RF-EMF exposures from cell
effects in healthy participants (N=47) of
ciesrecordedinneuronaltissueandcould
phones have reported increases,7,8 de-
acute cell phone exposures on brain glu-
interfere with neuronal activity.4 Ther-
creases and increases,9,10 or no changes11
cose metabolism, measured using PET
mal effects from RF-EMFs have also been
in CBF. The discrepancies among these
with injection of (18F)fluorodeoxyglu-
invoked as a mechanism that could affect
imaging studies likely reflect their rela-
cose (18FDG). Brain glucose metabolic
neuronal activity, although temperature
tively small sample sizes (9-14 partici-
changes produced by current cell phone
pants), and the potential confounding of
Author Affiliations: National Institute on Drug Abuse,
technology are likely minimal.5 Studies
Bethesda, Maryland (Dr Volkow); National Institute on
CBF measures reflecting vascular rather
Alcohol Abuse and Alcoholism, Bethesda (Drs Volkow,
performed in humans to investigate the
than neuronal signals.12-14 This highlights
Tomasi, and Telang and Mr Wong); and Medical Depart-
effects of RF-EMF exposures from cell
ment,BrookhavenNationalLaboratory,Upton,NewYork
the need for studies to document whether
(Drs Wang, Vaska, Fowler, and Logan and Mr Alexoff ).
RF-EMFs from cell phone use affects
Corresponding Author: Nora D. Volkow, MD, National
For editorial comment see p 828.
InstituteonDrugAbuse,6001ExecutiveBlvd,Room5274,
brain function in humans.
Bethesda, MD 20892 (nvolkow@nida.nih.gov).
808 JAMA, February 23, 2011—Vol 305, No. 8 (Reprinted)
©2011 American Medical Association. All rights reserved.

---

CELL PHONE SIGNALS AND BRAIN GLUCOSE METABOLISM
activity is a more proximal marker of
injection under resting conditions. For
Table 1. Characteristics and Cellular
neuronal activity than measures of CBF,
both scans 2 cell phones, one placed on
Telephone Histories of Participants (N = 47)
which reflects vascular as well as neu-
the left ear and one on the right, were
Characteristic
No. (%)
ronal components.15 Also, because brain
used to avoid confounding effects from
Age, mean (SD), y
31 (9)
glucose metabolic measures obtained
the expectation of a signal from the side
Sex
with 18FDG reflect the averaged brain ac-
of the brain at which the cell phone was
Men
23 (48.9)
tivity occurring over a 30-minute pe-
located. For one of the days both cell
Women
24 (51.1)
riod,16 this method allowed assessment
phones were deactivated (“off” condi-
Body mass index, mean (SD)a
26 (3)
of the cumulative effects of cell phone ex-
tion). For the other day the right cell
Handedness
Right-handed
43 (91.5)
posure on resting brain metabolism. Be-
phone was on (activated but muted to
Left-handed
4 (8.5)
cause exposure to RF-EMFs from cell
avoid confounding from auditory stimu-
Education mean (SD), y
14 (2)
phones is well localized and is highest
lation) and the left cell phone was off
Cell phone use, mean
1500 (1850)
in brain regions closest to the antenna,2
(“on” condition). For the on condition
(SD) [range], min/mo
[15-9000]
we hypothesized that the effects on brain
the cell phone was receiving a call (from
Ear favored for use
Right
38 (80.9)
metabolism would be greatest in infe-
a recorded text), although the sound was
Left
9 (19.1)
rior and anterior brain regions, the re-
muted. The order of conditions was ran-
a Calculated as weight in kilograms divided by height in me-
gions that would be exposed to the high-
domly assigned, and participants were
ters squared.
est RF-EMF amplitude for the cell phone
blinded to the condition. The mean time
model used in this study.
between the first and the second study
the head from a distance of 3 feet. The
was 5 (SD, 3) days.
cellular band was active, with a fre-
METHODS
Two Samsung model SCH-U310 cell
quency of 837.8 MHz. This frequency
Participants
phones, capable of transmitting at either
was monitored with a resolution band-
The study was conducted at Brookhaven
cellular or personal communications ser-
width of 1 MHz. Activation of the cell
National Laboratory from January 1,
vice frequency bands with code divi-
phone for the experimental period was
2009, through December 31, 2009, and
sion multiple access modulation, were
also corroborated with the records ob-
was approved by the local institutional
used for each study. The maximum spe-
tained from the cell phone company. For
review board (Committee on Research
cific absorption rate in the head for this
1 participant the cell phone signal was
Involving Human Subjects, Stony Brook
cell phone model corresponds to 0.901
interrupted at the time of 18FDG injec-
University). We enrolled 48 healthy par-
W/kg. Cell phones were placed over each
tion; this participant’s data were not in-
ticipants recruited from advertisements
ear with microphones directed toward
cluded in the analysis.
in local newspapers and screened for ab-
the participant’s mouth and were se-
sence of medical, psychiatric, or neuro-
cured to the head using a muffler that did
PET Scanning
logic diseases. Special attention was given
not interfere with the lower part of the
In preparation for the study, participants
to ensure that participants did not abuse
cell phone, where the antenna is lo-
had 2 venous catheters placed, one in the
addictive substances (including alco-
cated. Activation of the right cell phone
antecubital vein for radiotracer injection
hol and nicotine), and urine toxicology
was started 20 minutes prior to 18FDG
and the other in a superficial vein on the
studies were performed prior to the
injection and maintained for 30 min-
dorsal surface of the hand for sampling
imaging sessions to ensure lack of psy-
utes afterward to correspond with the
of arterialized blood. Arterialization was
choactive drug use. For technical rea-
18FDG uptake period. During the 50-
achieved by warming the hand to 44°C.
sons, data from one of the participants
minute period participants sat on a com-
The participants were injected with
could not be used (see below). TABLE 1
fortable chair in a quiet, dimly lit room
18FDG (148-222 MBq [to convert to
provides demographic characteristics and
and with their eyes open, with a nurse
millicuries, divide by 37]) and asked
cell phone usage histories of the 47 par-
present to ensure that they kept their eyes
to refrain from moving or speaking
ticipants whose data were used in the
open and did not fall asleep.
during the 30-minute 18FDG uptake pe-
analysis. Participants each received $250
The RF-EMF emissions were re-
riod. At the end of the sessions, the cell
for their participation in the study ($200
corded once before the call (back-
phones were removed and the partici-
for PET scans [$100 per scan] plus $50
ground) and every 5 minutes during the
pants were positioned in the PET scan-
for the physical examination and labo-
stimulation period to ensure that the call
neraspreviouslydescribed.17Participants
ratory work). All participants provided
was not terminated. This was accom-
were scanned with a whole-body tomo-
written informed consent after receiv-
plished with a handheld spectrum ana-
graph (ECAT HR ; Siemens/CTI,
ing a complete description of the study.
lyzer (model FSH6; Rohde & Schwarz,
Munich, Germany), with a resolution of
Munich, Germany) connected to a cel-
4.6
4.6
4.2 mm3 as measured by Na-
Experimental Conditions
lular wide-band log periodic direc-
tional Electrical Manufacturers Associa-
All participants had 2 scans performed
tional antenna (model 304411; Wilson
tion protocols. Emission scans were
on separate days using PET with 18FDG
Electronics, St. George, Utah) aimed at
started 35 minutes after 18FDG injection
©2011 American Medical Association. All rights reserved.
(Reprinted) JAMA, February 23, 2011—Vol 305, No. 8 809

---

CELL PHONE SIGNALS AND BRAIN GLUCOSE METABOLISM
on glucose metabolism would occur in
Figure 1. Amplitude of the Electric Field Emitted by the Right Cellular Telephone Antenna
regions close to the antenna and that the
Rendered on the Surface of the Human Brain
regions far from the antenna would show
R I G H T H E M I S P H E R E
L E F T H E M I S P H E R E
no effects. Therefore, the corrections for
Lateral view
Lateral view
multiple comparisons were restricted to
brain regions in which E(r) was higher
than 50% of the maximum field value,
E0, in the brain (E0/2
E(r)
E0)
(Figure 1). Thus, the Bonferroni method
with a searching volume (Sv) of 201.3
cm3 (Sv=25 161 voxels) was used to cor-
rect cluster-level P values for multiple
comparisons as a function of the cluster
Amplitude of electric field, E(r)
volume (Cv) (Pcorr=P Sv/Cv). Clusters
with at least 1000 voxels (Cv
8 cm3)
0
E (maximum)
0
Medial view
Medial view
and P
.05 (corrected for multiple com-
Boundary of clusters proximal
parisons) were considered significant.
to antenna (E /2<E(r)<E )
0
0
A simple model assuming a linear re-
lationship between cell phone–related
increases in metabolism ( 18FDG; av-
erage across participants) and E was
used. The paired values ( 18FDGi, Ei)
from all voxels that were statistically sig-
nificant in the SPM2 t test analyses con-
trasting on vs off conditions within Sv
were sorted by E, clustered in groups
of 50 voxels, and averaged. These clus-
E0 indicates maximal field value. Clusters proximal to the antenna are inferior to the red dashed line. Images
ters were treated as independent. The
created using the freeware Computerized Anatomical Reconstruction and Editing Toolkit (CARET) version 5.0
(http://brainvis.wustl.edu/wiki/index.php/Caret:About).
Pearson linear correlation factor, R, was
used to assess the linear relationship be-
tween 18FDG and E in Interactive Data
and lasted 20 minutes. Transmission
tion Solutions, Boulder, Colorado) using
Language version 6.0.
scans were performed simultaneously.
the far-field approximation, E(r) ~ ||r-
The sample-size calculation was based
r
on our preliminary study of the effect of
0||−3, of a dipole field (FIGURE 1).
Radiofrequency Field
low-frequency magnetic field gradients
The average position of the antenna in
Image Analysis
in glucose metabolism,19 which demon-
the stereotactic space of the Montreal
The data were analyzed using statistical
strated metabolic differences between
Neurological Institute (r0) (r0=21 [SD,
parametric mapping (SPM) in the SPM2
stimulation and sham conditions with
10] mm for x [left to right], 30 [SD, 11]
mappingpackage(WelcomeDepartment
effect size (ratio between the mean differ-
mm for y [anterior to posterior], −160
of Cognitive Neurology, London, United
ence and the pooled standard deviation)
[SD, 7] mm for z [superior to inferior])
Kingdom).18 The SPM analyses were per-
between 0.65 and 0.80. The minimal im-
was determined for 21 participants using
formed on the absolute as well as the nor-
portant difference in glucose metabolism
calibrated orthogonal photography that
malized (to whole-brain metabolism)
used to determine the sample size was 1
registered orthogonal views (front and
metabolic images. For this purpose, the
µmol/100 g per minute. For such effect
sides) of the cell phone positions on the
images were spatially normalized using
sizes, to achieve a power of at least 80%
participant’s head. The positions of the
the SPM2 PET template and a 2-mm3 2-
using the independent-samples t test with
eyes were used as landmarks to deter-
mm3
2-mm3 voxel size and were sub-
a significance level of .05, at least 40 par-
mine r0 with the aid of the standard brain
sequently smoothed with an 8-mm iso-
ticipants were needed.
template (ch2.nii) provided in MRI-
tropic Gaussian kernel. Voxel-wise
cron (available at http://www.sph.sc.edu
paired t tests were used to assess regional
RESULTS
/comd/rorden/mricron/). The relative
changes in glucose metabolism.
Whole-brain glucose metabolism did not
amplitude of the cell phone’s electric
Because the electric field, E(r), pro-
differ between conditions, which for the
field, E(r), at every position in the brain,
duced by the cell phone decreases rap-
off condition corresponded to 41.2 µmol/
r, was computed in Interactive Data Lan-
idly with distance to the antenna, we hy-
100 g per minute (95% confidence inter-
guage version 6.0 (ITT Visual Informa-
pothesized that the effects of cell phones
val [CI], 39.5-42.8) and for the on con-
810 JAMA, February 23, 2011—Vol 305, No. 8 (Reprinted)
©2011 American Medical Association. All rights reserved.

---

CELL PHONE SIGNALS AND BRAIN GLUCOSE METABOLISM
dition to 41.7 µmol/100 g per minute
Figure 2. Brain Glucose Metabolic Images Showing Axial Planes at the Level of the
(95% CI, 40.1-43.3). However, there were
Orbitofrontal Cortex
significant regional effects. Specifically,
the SPM comparisons14 on the absolute
Cell phone on
Cell phone off
metabolic measures showed significant
L
R
L
R
increases (35.7 vs 33.3 µmol/100 g per
minute for the on vs off conditions,
respectively; mean difference, 2.4 [95%
CI, 0.67-4.2]; P=.004) in a region that
included the right orbitofrontal cortex
(BA 11/47) and the lower part of the right
superior temporal gyrus (BA 38)
(FIGURE 2 and TABLE 2). No areas
showed decreases. Similar results were
obtained for the SPM analysis of the nor-
malized metabolic images (normal-
ized to whole-brain glucose metabo-
lism), which also showed significant
increases (1.048 vs 0.997 for the on vs
off conditions, respectively; mean dif-
Rate of glucose metabolism,
µmol/100 g per min
ference, 0.051 [95% CI, 0.017-0.091];
P
.001) in a region that included right
0
60
orbitofrontal cortex and right superior
Images are from a single participant representative of the study population. Glucose metabolism in right or-
temporal gyrus (BA 38) (Figure 2).
bitofrontal cortex (arrowhead) was higher for the “on” than for the “off” condition (see “Methods” for de-
The regression analysis between cell
scription of conditions).
phone–related increases in metabolism
( 18FDG) and E revealed a significant
cur in brain regions with the highest RF-
RF-EMF exposure, whereas findings
positive correlation both for the absolute
EMFexposures.7-10Moreover,oneofthese
from other studies have shown de-
metabolic measures (R=0.95, P
.001)
studies reported CBF decreases in the re-
creases in regions with the highest RF-
and the normalized metabolic measures
gion with maximal RF-EMF exposure.10
EMF exposures, increases in regions far
(R=0.89, P
.001) (FIGURE 3). This in-
These discrepancies are likely to reflect,
from the antenna, or both. However, the
dicates that the regions expected to have
among others, the methods used, particu-
increases in frontal CBF previously re-
the greater absorption of RF-EMFs from
larly because the 18FDG method is opti-
ported with acute cell phone exposure
the cell phone exposure were the ones
mal for detecting long-lasting effects (30
possibly could reflect a downstream effect
that showed the larger increases in glu-
minutes) in brain activity, whereas CBF
of connections with the regions that had
cose metabolism.
measures reflect activity over 60 seconds.
the highest RF-EMF exposures.
In this respect, this study is an example
The linear association between cell
CONCLUSIONS
of the value of the 18FDG method for de-
phone–related increases in metabolism
These results provide evidence that the
tecting cumulative effects in brain activ-
( 18FDG) and E suggests that the meta-
human brain is sensitive to the effects of
ity that may not be observed when using
bolic increases are secondary to the
RF-EMFs from acute cell phone expo-
more transient measures of activity. Dis-
absorption of RF-EMFs from cell phone
sures. The findings of increased metabo-
crepancies also could reflect uncoupling
exposures. The mechanisms by which
lism in regions closest to the antenna dur-
between CBF and metabolism.12-14 More-
RF-EMFs from cell phones could affect
ing acute cell phone exposure suggest
over, the relatively large sample size
brain glucose metabolism are unclear.
that brain absorption of RF-EMFs may
(n=47) improved our ability to detect
However, based on findings from in vivo
enhance the excitability of brain tissue.
small effects that may have been missed
animal and in vitro experiments, it has
This interpretation is supported by a re-
inpriorstudieswithsmallersamplesizes.11
been hypothesized that this could reflect
port of enhanced cortical excitability to
The experimental setup also differed
effects of RF-EMF exposure on neuronal
short transcranial magnetic stimulation
from prior studies that used cell phones
activity mediated by changes in cell mem-
pulses (1 msec) following 40-minute RF-
for which the antenna was closest to su-
brane permeability, calcium efflux, cell
EMF exposures.20
perior and middle temporal cortices.21
excitability, and/or neurotransmitter re-
Although increases in frontal CBF dur-
However, this is unlikely to have ac-
lease.4 A thermal effect of cell phones on
ing acute cell phone exposure had been
counted for the differences in results, be-
thebrainhasalsobeenproposed,22butthis
previouslyreportedby2independentPET
cause the findings in this study show
isunlikelytocontributetofunctionalbrain
laboratories, such increases did not oc-
increases in the region with maximal
changes.5 Disruption of the blood-brain
©2011 American Medical Association. All rights reserved.
(Reprinted) JAMA, February 23, 2011—Vol 305, No. 8 811

---

CELL PHONE SIGNALS AND BRAIN GLUCOSE METABOLISM
Table 2. Statistical Parametric Mapping For Brain Regions Showing Higher Glucose Metabolism With Cellular Telephone On Than Off
Region Coordinates,
mmb
On vs Off,
Z Score,
Mean Difference
Brain Region
Volumea
Brodmann Area
x
y
z
On vs Off
P
c
corr
(95% CI)
Absolute glucose metabolism
Right inferior frontal
47
18
23
−18
2.7
Right superior temporal
2649
38
24
12
−37
2.6
.05
2.4 (0.67-4.2)d
Right middle frontal
11
23
38
−15
2.6
Normalized glucose metabolism
Right superior temporal
38
27
2
−35
3.1
Right inferior frontal
2910
47
16
27
−16
3.1
.05
7.8 (2.7-12.9)d
Right middle frontal
11
23
38
−15
3.1
Abbreviation: CI, confidence interval.
a No. of voxels. One voxel=0.008 mm3.
b Coordinates on the Montreal Neurological Institute stereotactic space corresponding to distance (in mm) for x (left to right), y (anterior to posterior), and z (superior to inferior).
c See “Methods” for details of calculation of Bonferroni-corrected P value.
d Values for absolute metabolism reported in µmol/100 g per minute; those for normalized metabolism reported as percentages.
barrier has also been invoked as a poten-
Figure 3. Measures of Absolute and Normalized Glucose Metabolism and Correlation Between
tial mechanism by which RF-EMFs from
Estimated Electromagnetic Field Amplitudes and Increases in Measures (N=47 Participants)
cell phone exposure could affect brain ac-
tivity.23 A recent clinical study reported
A Absolute glucose metabolism
B Normalized glucose metabolism
alterations in a peripheral biomarker of
45
1.4
Cell phone
Cell phone
blood-brain barrier integrity (transthyre-
On
On
tin) after cell phone exposure, but the sig-
Off
Off
40
1.2
nificance of this finding is unclear.24
The increases in regional metabolism
35
inducedbyRF-EMFs(approximately7%)
1.0
are similar in magnitude to those reported
30
µmol/100 g per min
Region/Whole Brain
aftersuprathresholdtranscranialmagnetic
25
0.8
stimulation of the sensorimotor cortex
0.5
0.6
0.7
0.8
0.9
1.0
0.5
0.6
0.7
0.8
0.9
1.0
(7%-8%).25 However, these increases are
E /E
E /E
0
0
much smaller than the increases after vi-
sual stimulation reported by most stud-
C Change in absolute glucose metabolism
D Change in normalized glucose metabolism
ies (range, 6%-51%).26 The large differ-
8
10
ence in the magnitude of regional glucose
8
f
f
metabolicincreasesislikelytoreflectmul-
6
6
tiple factors, including differences in gly-
om Of
4
om Of
4
colytic rate between brain regions,27 the
2
duration of the stimulation (transient
2
0
stimulationincreasesglucosemetabolism
–2
% Change Fr
0
% Change Fr
morethancontinuousstimulation26),and
–4
the characteristics of the stimulation
–2
–6
0.5
0.6
0.7
0.8
0.9
1.0
0.5
0.6
0.7
0.8
0.9
1.0
used.28 Indeed, whereas resting glucose
E /E
E /E
0
0
metabolism is predominantly supported
by glucose oxidation ( 90%), with acute
A and B, Mean measures of absolute glucose metabolism (µmol/100 g per minute) and normalized glucose
visual stimulation the large increases in
metabolism (region/whole brain; units cancel) in regions with increased metabolism during “on” vs “off”
conditions (see “Methods” for description of conditions) in the brain area within the spherical constraint,
glucose metabolism appear to reflect pre-
E0/2 E(r) E0 (where E0 indicates maximal field value and E(r) indicates amplitude of the theoretical elec-
dominantly aerobic glycolysis,29 which is
tromagnetic field) and the E(r) emitted by the antenna of the right cellular telephone. Absolute=40 clusters;
used for purposes other than energy ex-
2000 voxels were activated within searching volume and grouped into clusters of 50 voxels each; normal-
ized = 48 clusters; 2400 voxels were activated within searching volume and grouped into clusters of 50
penditures, and actual energy utilization
voxels each. Range of variability (95% confidence interval [CI]): 9-21 µmol/100 g per minute (panel A)
is estimated to be 8% at most.13
and 0.29-0.57 (panel B). C and D, Regression lines between cell phone–related increases in absolute and
normalized glucose metabolism (both expressed as % change from the off condition) in brain regions
Concern has been raised by the pos-
within the spherical constraint, E0/2 E(r) E0, and the theoretical electric field, E(r), emitted by the antenna
sibility that RF-EMFs emitted by cell
of the right cell phone. Increases significantly correlated with estimated electromagnetic field amplitudes
(absolute: R = 0.95, P
.001; normalized: R = 0.89, P
.001). Data markers indicate mean metabolic mea-
phones may induce brain cancer.30 Epi-
sures; error bars, 95% CIs. Linear regression lines were fitted to the data using Interactive Data Language
demiologic studies assessing the rela-
version 6.0.
tionship between cell phone use and rates
812 JAMA, February 23, 2011—Vol 305, No. 8 (Reprinted)
©2011 American Medical Association. All rights reserved.

---

CELL PHONE SIGNALS AND BRAIN GLUCOSE METABOLISM
of brain cancers are inconclusive; some
Conflict of Interest Disclosures: All authors have com-
15. Iadecola C, Nedergaard M. Glial regulation of the
pleted and submitted the ICMJE Form for Disclosure of
cerebral microvasculature. Nat Neurosci. 2007;
report an association,31-33 whereas oth-
Potential Conflicts of Interest and none were reported.
10(11):1369-1376.
ers do not.34-36 Results of this study pro-
Funding/Support: This study was carried out at
16. Sokoloff L, Reivich M, Kennedy C, et al. The
[14C]deoxyglucose method for the measurement of
vide evidence that acute cell phone ex-
Brookhaven National Laboratory (BNL) and was sup-
ported by the Intramural Research Program of the Na-
local cerebral glucose utilization. J Neurochem. 1977;
posure affects brain metabolic activity.
tional Institutes of Health (NIH) and by infrastructure
28(5):897-916.
However, these results provide no infor-
support from the Department of Energy.
17. Wang G-J, Volkow ND, Roque CT, et al. Func-
Role of Sponsor: The funding agencies had no role in the
tional importance of ventricular enlargement and cor-
mation as to their relevance regarding po-
design and conduct of the study; the collection, manage-
tical atrophy in healthy subjects and alcoholics as as-
tential carcinogenic effects (or lack of
ment,analysis,andinterpretationofthedata;ortheprepa-
sessed with PET, MR imaging, and neuropsychologic
ration, review, or approval of the manuscript.
testing. Radiology. 1993;186(1):59-65.
such effects) from chronic cell phone use.
Additional Contributions: We are grateful to BNL em-
18. Friston KJ, Holmes AP, Worsley KJ, et al. Statis-
Limitations of this study include that
ployees Donald Warner, AA, for positron emission to-
tical parametric maps in functional imaging. Hum Brain
Mapp. 1995;2:189-210.
it is not possible to ascertain whether the
mography operations; David Schlyer, PhD, and Michael
Schueller, PhD, for cyclotron operations; Pauline Carter,
19. Volkow ND, Tomasi D, Wang GJ, et al. Effects
findings pertain to potential harmful ef-
RN, and Barbara Hubbard, RN, for nursing care; Payton
of low-field magnetic stimulation on brain glucose
metabolism. Neuroimage. 2010;51(2):623-628.
fects of RF-EMF exposures or only docu-
King, BS, for plasma analysis; and Lisa Muench, MS, You-
20. Ferreri F, Curcio G, Pasqualetti P, et al. Mobile
wen Xu, MS, and Colleen Shea, MS, for radiotracer prepa-
ment that the brain is affected by these ex-
phone emissions and human brain excitability. Ann
ration; and to NIH employees Karen Appelskog-Torres,
Neurol. 2006;60(2):188-196.
posures. Also, this study does not provide
AA, for protocol coordination; Millard Jayne, RN, for sub-
21. Cardis E, Deltour I, Mann S, et al. Distribution of
ject recruitment and nursing care; and Linda Thomas, MS,
anunderstandingofthemechanism(s)by
RF energy emitted by mobile phones in anatomical
for editorial assistance. We also thank the individuals who
structures of the brain. Phys Med Biol. 2008;53
which RF-EMF exposures increase brain
volunteered for these studies. None of the individuals ac-
(11):2771-2783.
knowledged were compensated in addition to their sala-
metabolism, and although we interpret
22. Cotgreave IA. Biological stress responses to ra-
ries for their contributions.
dio frequency electromagnetic radiation. Arch Bio-
these exposures as indicators of neuronal
chem Biophys. 2005;435(1):227-240.
excitation, further studies are necessary
REFERENCES
23. Nittby H, Grafström G, Eberhardt JL, et al. Radio-
frequency and extremely low-frequency electromag-
to corroborate this. Lastly, this model as-
1. Dubey RB, Hanmandlu M, Gupta SK. Risk of brain
netic field effects on the blood-brain barrier. Electro-
sumes a linear relationship between the
tumors from wireless phone use. J Comput Assist
magn Biol Med. 2008;27(2):103-126.
Tomogr. 2010;34(6):799-807.
24. Söderqvist F, Carlberg M, Hansson Mild K, Hardell
amplitude of the radiofrequency field and
2. Schönborn F, Burkhardt M, Kuster N. Differences
L. Exposure to an 890-MHz mobile phone-like signal
its effects in neuronal tissue, but we can-
in energy absorption between heads of adults and chil-
and serum levels of S100B and transthyretin in
not rule out the possibility that this rela-
dren in the near field of sources. Health Phys. 1998;
volunteers. Toxicol Lett. 2009;189(1):63-66.
74(2):160-168.
25. Siebner HR, Peller M, Bartenstein P, et al. Acti-
tionship could be nonlinear.
3. Kleinlogel H, Dierks T, Koenig T, Lehmann H, Minder
vation of frontal premotor areas during suprathresh-
In summary, this study provides evi-
A, Berz R. Effects of weak mobile phone–
old transcranial magnetic stimulation of the left
electromagnetic fields (GSM, UMTS) on event re-
primary sensorimotor cortex: a glucose metabolic
dence that in humans RF-EMF exposure
l a t e d p o t e n t i a l s a n d c o g n i t i v e f u n c t i o n s .
PET study. Hum Brain Mapp. 2001;12(3):157-
fromcellphoneuseaffectsbrainfunction,
Bioelectromagnetics. 2008;29(6):488-497.
167.
4. Hyland GJ. Physics and biology of mobile telephony.
26. Vlassenko AG, Rundle MM, Mintun MA. Hu-
asshownbytheregionalincreasesinmeta-
Lancet. 2000;356(9244):1833-1836.
man brain glucose metabolism may evolve during
bolic activity. It also documents that the
5. Wainwright P. Thermal effects of radiation from
activation. Neuroimage. 2006;33(4):1036-1041.
cellular telephones. Phys Med Biol. 2000;45(8):
27. Vaishnavi SN, Vlassenko AG, Rundle MM, et al.
observed effects were greatest in brain re-
2363-2372.
Regional aerobic glycolysis in the human brain. Proc
gions that had the highest amplitude of
6. van Rongen E, Croft R, Juutilainen J, et al. Effects
Natl Acad Sci U S A. 2010;107(41):17757-17762.
of radiofrequency electromagnetic fields on the hu-
28. Sanganahalli BG, Herman P, Hyder F. Frequency-
RF-EMF emissions (for the specific cell
man nervous system. J Toxicol Environ Health B Crit
dependent tactile responses in rat brain measured by
phones used in this study and their po-
Rev. 2009;12(8):572-597.
functional MRI. NMR Biomed. 2008;21(4):410-
7. Huber R, Treyer V, Borbe´ly AA, et al. Electromag-
416.
sition relative to the head when in use),
netic fields, such as those from mobile phones, alter
29. Blomqvist G, Seitz RJ, Sjögren I, et al. Regional
which suggests that the metabolic in-
regional cerebral blood flow and sleep and waking EEG.
cerebral oxidative and total glucose consumption dur-
J Sleep Res. 2002;11(4):289-295.
ing rest and activation studied with positron emission
creases are secondary to the absorption
8. Huber R, Treyer V, Schuderer J, et al. Exposure to
tomography. Acta Physiol Scand. 1994;151(1):
of RF-EMF energy emitted by the cell
pulse-modulated radio frequency electromagnetic fields
29-43.
affects regional cerebral blood flow. Eur J Neurosci.
30. Yakymenko I, Sidorik E. Risks of carcinogenesis
phone. Further studies are needed to as-
2005;21(4):1000-1006.
from electromagnetic radiation of mobile telephony
sess if these effects could have potential
9. Haarala C, Aalto S, Hautzel H, et al. Effects of a
devices. Exp Oncol. 2010;32(2):54-60.
902 MHz mobile phone on cerebral blood flow in
31. Lehrer S, Green S, Stock RG. Association be-
long-term harmful consequences.
humans. Neuroreport. 2003;14(16):2019-2023.
tween number of cell phone contracts and brain tu-
10. Aalto S, Haarala C, Bruck A, et al. Mobile phone
mor incidence in nineteen U.S. States. J Neurooncol.
Author Contributions: Drs Volkow and Tomasi had
affects cerebral blood flow in humans. J Cereb Blood
2011;101(3):505-507.
full access to all of the data in the study and take re-
Flow Metab. 2006;26(7):885-890.
32. Hardell L, Carlberg M. Mobile phones, cordless
sponsibility for the integrity of the data and the ac-
11. Mizuno Y, Moriguchi Y, Hikage T, et al. Effects
phones and the risk for brain tumours. Int J Oncol.
curacy of the data analysis.
of W-CDMA 1950 MHz EMF emitted by mobile
2009;35(1):5-17.
Study concept and design: Volkow.
phones on regional cerebral blood flow in humans.
33. Myung SK, Ju W, McDonnell DD, et al. Mobile
Acquisition of data: Wang,Vaska,Telang,Alexoff,Wong.
Bioelectromagnetics. 2009;30(7):536-544.
phone use and risk of tumors: a meta-analysis. J Clin
Analysis and interpretation of data: Volkow, Tomasi,
12. Sirotin YB, Das A. Anticipatory haemodynamic sig-
Oncol. 2009;27(33):5565-5572.
Vaska, Fowler, Telang, Logan.
nals in sensory cortex not predicted by local neuronal
34. Inskip PD, Tarone RE, Hatch EE, et al. Cellular-
Drafting of the manuscript: Volkow, Wong.
activity. Nature. 2009;457(7228):475-479.
telephone use and brain tumors. N Engl J Med. 2001;
Critical revision of the manuscript for important in-
13. Fox PT, Raichle ME, Mintun MA, Dence C. Non-
344(2):79-86.
tellectual content: Volkow, Tomasi, Wang, Vaska,
oxidative glucose consumption during focal physi-
35. INTERPHONE Study Group. Brain tumour risk in
Fowler, Telang, Alexoff, Logan.
ologic neural activity. Science. 1988;241(4864):
relation to mobile telephone use. Int J Epidemiol. 2010;
Statistical analysis: Tomasi.
462-464.
39(3):675-694.
Obtained funding: Volkow, Fowler.
14. Devor A, Hillman EM, Tian P, et al. Stimulus-
36. Inskip PD, Hoover RN, Devesa SS. Brain cancer
Administrative, technical, or material support: Wang,
induced changes in blood flow and 2-deoxyglucose
incidence trends in relation to cellular telephone use
Fowler, Telang, Alexoff, Wong.
uptake dissociate in ipsilateral somatosensory cortex.
in the United States. Neuro Oncol. 2010;12(11):
Study supervision: Wang, Fowler.
J Neurosci. 2008;28(53):14347-14357.
1147-1151.
©2011 American Medical Association. All rights reserved.
(Reprinted) JAMA, February 23, 2011—Vol 305, No. 8 813

---