вход по аккаунту


Caffeine and brain development in very preterm infants.

код для вставкиСкачать
Caffeine and Brain Development in Very
Preterm Infants
Lex W. Doyle, MD,1,2,3,7 Jeanie Cheong, MD,1,2 Rod W. Hunt, PhD,5,7
Katherine J. Lee, PhD,7 Deanne K. Thompson, BSc,6,7 Peter G. Davis, MD,1,7
Sandra Rees, PhD,4 Peter J. Anderson, PhD,2,7 and Terrie E. Inder, MD7,8
Objective: Caffeine improves neurological outcome in very preterm infants, but the mechanisms responsible for this
neurological benefit are unknown. The objective of this study was to assess whether caffeine influenced brain macroor microstructural development in preterm infants.
Methods: Seventy preterm infants <1,251 g birthweight randomly allocated to either caffeine (n ¼ 33) or placebo
(n ¼ 37) underwent brain magnetic resonance imaging (MRI) at term-equivalent age; white and gray matter
abnormalities were qualitatively scored, global and regional brain volumes were measured, and white matter
microstructure was evaluated using diffusion-weighted imaging.
Results: There were no significant differences between the groups in the extent of white matter or gray matter
abnormality, or in global or regional brain volumes. In contrast, although only available in 28 children, caffeine
exposure was associated with reductions in the apparent diffusion coefficient, and radial and axial diffusivity with the
greatest impact in the superior brain regions. The alterations in diffusion measures were not mediated by lowering
the rate of lung injury, known as bronchopulmonary dysplasia.
Interpretation: These diffusion changes are consistent with improved white matter microstructural development in
preterm infants who received caffeine.
ANN NEUROL 2010;68:734–742
affeine therapy, widely used to reduce apnea
in preterm infants, substantially improves both
short- and long-term outcomes in infants of birthweight
500–1,250 g.1,2 In the recently completed Caffeine in
Apnea of Prematurity (CAP) study, the rate of lung
injury, called bronchopulmonary dysplasia (BPD), was
lower with caffeine treatment started within the first 10
days of life compared with placebo.1 Of greater importance, at 18 months’ corrected age cerebral palsy and developmental delay were both less common, and the rate
of survival free of major neurological disability was substantially higher with caffeine therapy.2
Caffeine exposure in immature rodent models and
cell culture can be toxic3,4 or protective5–9 to neurons
and glia. However, caffeine effects in human newborns
remain to be determined.
In one of the centers participating in the CAP
study there was a simultaneous parallel study of cerebral
injury and development, known as the Victorian Infant
Brain Study (VIBeS). The longitudinal VIBeS study
involved magnetic resonance imaging (MRI) at term
equivalent age with follow-up into childhood. As some
infants participated in both studies, an opportunity
existed to investigate the effects of caffeine treatment on
cerebral structure within the context of a randomized
controlled trial.
The objective of this study was to determine the
relationship between caffeine therapy and measures of
cerebral injury and development using MRI. It was
hypothesized that caffeine treatment would have a beneficial effect in reducing white matter injury and improving
cerebral white matter development in preterm infants.
The Royal Women’s Hospital in Melbourne contributed 199 of
the 2,006 infants enrolled in the CAP study. Within the CAP
View this article online at DOI: 10.1002/ana.22098
Received Jan 25, 2010, and in revised form May 5, 2010. Accepted for publication May 19, 2010.
Address correspondence to Professor Lex Doyle, 7th Floor Research Precinct, Royal Women’s Hospital, Locked Bag 300, Parkville, Victoria, Australia,
3052. E-mail:
From the 1Royal Women’s Hospital, Department of 2Obstetrics and Gynaecology, 3Paediatrics, 4Anatomy and Cell Biology, University of Melbourne,
Australia; 5The Royal Children’s Hospital; 6Howard Florey Institute; 7Murdoch Children’s Research Institute, Melbourne, Australia; 8Washington University,
St. Louis, Missouri, USA.
C 2010 American Neurological Association
734 V
Doyle et al: Caffeine and the Brain
study, preterm infants of birthweight 500–1,250 g considered
suitable for caffeine therapy within the first 10 days were randomly allocated to either caffeine citrate with a loading dose of
20 mg/kg and maintenance dose of 5–10 mg/kg/day, or an
equal volume of saline placebo. The study drug was discontinued permanently at the discretion of the local clinicians. However, it was recommended to continue therapy with the study
drug until the infant had tolerated at least 5 consecutive days
without positive airway pressure. Study infants received their
first doses of caffeine or placebo at a median postnatal age of 3
days and a median postmenstrual age of 28 weeks. The median
duration of treatment was 37 days in the caffeine group and 36
days in the placebo group; the median age at stopping treatment was 34 weeks’ postmenstrual age. Other details of the
CAP study are reported elsewhere.1,2
Seventy infants cared for at the Royal Women’s Hospital
and enrolled in the CAP study were also in the VIBeS study,
and hence underwent MRI scans at 38–42 weeks’ postmenstrual
age. No infants were on study medication at the time of the
MRI. These 70 children are the subjects of this article. Ethical
permission was granted by the Royal Women’s Hospital
Research and Ethics Committees and written informed parental
consent was necessary for participation in both studies.
Magnetic Resonance Methods
MRI was performed without sedation between 38 to 42 weeks’
corrected gestational age using a 1.5 T General Electric Signa
System (GE Medical Systems) with the following sequences: 1)
a 3D Fourier transform spoiled gradient recalled (SPGR)
sequence (1.2 mm coronal slices, flip angle 45 , repetition time
(TR) 35 ms, echo time (TE) 9 ms, field of view (FOV) 21 15 cm2, matrix 256 192); 2) a double-echo (proton density
and T2-weighted) spin-echo sequence (dual-echo) (2 mm axial
slices, TR 4,000 ms, TE 60 and 160 ms, FOV 22 16 cm2,
matrix 256 192, interpolated 512 512, interleaved acquisition); and 3) a linescan diffusion sequence (4–6 mm axial slices
with a 0.5–1 mm gap, TR 2139 ms, TE 78 ms, FOV 22 cm,
matrix 128 128, 2 images at b ¼ 5 s/mm2, 6 images at b ¼
700 s/mm2). The diffusion gradients for b ¼ 700 s/mm2 were
oriented in six noncollinear directions. Acquisition time was 60
seconds per slice with 14–16 slices per brain.
The MR images were analyzed for the following: 1) qualitative white and gray matter abnormalities; 2) quantitative regional cerebral volumes including the hippocampi; 3) diffusion
tensor imaging (DTI) with measures of apparent diffusion coefficient (ADC), fractional anisotropy (FA), and radial and axial
diffusivity. Not all images were able to be analyzed for all variables because of the suboptimal quality of the images in some
infants. DTI acquisition only commenced in the latter part of
the VIBeS study; only 41 of the 70 CAP infants were enrolled
after DTI was added to the scanning protocol. The linescan
sequence was the last MR sequence obtained and was therefore
more vulnerable to movement artifact if the infant became
unsettled. In 11 infants after the DTI sequence was added, linescan was not even attempted because the infant became too
restless, and in two infants the DTI data were not able to be
November, 2010
processed due to excessive motion. Thus, only 28 of the 70
infants randomized into the CAP study had DTI data. All brain
imaging and analysis were performed independent of knowledge
of the infant’s exposure to caffeine.
Qualitative MR Analysis
Qualitative MRIs were classified according to degree of white
matter (WM) and gray matter (GM) abnormality, as previously
reported.10 WM abnormalities were categorized as normal/
mild, or moderate/severe white, and GM abnormalities were
categorized as normal or abnormal. Inter- and intrarater group
assignment was 94% or greater.
Quantitative MR Analysis
Quantitative volumetric MR analysis was undertaken by a single operator and the methods for total brain and hippocampal
segmentation have previously been described in detail.11,12 For
total brain segmentation, brain tissue was separated into myelinated and unmyelinated WM, cortical and deep nuclear
GM, and cerebrospinal fluid (CSF). Total brain volume
included all the GM and WM within the skull, including the
cerebellum, but excluded all CSF. Tissue segmentation was
completed on five infants blinded for 10 segmentations with
intraobserver reliability on tissue segmentation of CSF 0.99,
CGM 0.85, myelinated WM 0.73, unmyelinated WM 0.83,
and deep nuclear GM 0.61.
Diffusion Analysis
Line scan diffusion image acquisition and analysis was chosen for
these infants due to reduced impact of motion, distortions and
eddy current related artifacts, and reduced intrasubject variability.13 The disadvantage of line scan diffusion imaging was the
length of time required for acquisition (15–18 minutes per subject for 14–16 slices per infant). Line scan diffusion image analysis was performed offline using XPhase image analysis software
(SE Maier, Boston, MA). ADC, FA, axial (k1), and radial ([k2
þk3]/2) diffusivity were calculated.14,15 Parametric maps from
two slices were utilized for analysis. On a superior slice, just above
the lateral ventricles, six regions of interest (ROIs) were selected
and manually placed within each of the left and the right hemispheric WM (Fig. 1A). The inferior slice was taken at a level
through the basal ganglia and posterior limb of the internal capsule (PLIC) and ROIs were placed in the bilateral middle third of
the PLICs, and the anterior (frontal) and posterior (occipital)
WM regions (Fig. 1B). The size of the ROIs was uniform
between patients and a slice above and below the ROIs were
assessed to confirm that there was no CSF or GM that may contribute to partial voluming artifacts. All regions were placed on
the anisotropy map alongside the diffusivity and coregistered T2weighted images to optimize placement. All ROIs avoided any
areas of focal T1 or T2 signal abnormalities. All regions were
checked by two analysts and a senior investigator (T.I.). These
methods of ROI placement were developed in 2003 at the time
of commencement of data acquisition. Interobserver and intraobserver variability were assessed in 12 infants and were low, with
no statistical differences between observers (p ¼ 0.9) or between
of Neurology
Stata v. 10 (College Station, TX), with the mixed models fitted
using SAS v. 9.1 (Cary, NC).
Treatment allocation to caffeine or placebo within the
CAP study remained blinded to all clinicians, families, and
investigators, apart from the statisticians, in the current study.
FIGURE 1: Regions of interest (ROIs) for diffusion tensor
(DT) magnetic resonance imaging (MRI) in the superior (A)
and inferior (B) axial slices. (A) Image at the level just above
the superior margins of the lateral ventricles. ROIs in
anterior, central (sensorimotor), and posterior white matter.
(B) Image at the level of the basal ganglia and posterior
limbs of the internal capsule (PLIC). ROIs in anterior, central
(middle third of the PLIC), and posterior white matter.
test–retest (p ¼ 0.7). In addition, a subgroup of 10 infants had a
set of six larger ROIs (4-fold in voxel size) in addition to the
standard ROIs placed in the superior slice. There was high correlation (r ¼ 0.96) between the measures for each region between
the smaller and larger ROIs.
Statistical Analysis
Outcomes before discharge are presented as proportions and are
compared between groups using chi-square tests. WM and GM
injury are summarized as the median (interquartile range) for
the total score as well as the proportions for the categorized
scores and are compared between the groups using Wilcoxon
rank sum and chi-square tests, respectively. Crude means (SD)
are presented separately for the two groups for the rest of the
MRI data. Comparisons of segmented volumes between groups
were assessed in separate unadjusted linear regression models
for each volume/type of tissue. Hippocampal volumes were
compared combining the results from each hemisphere into a
single model after checking for an interaction between group
and hemisphere. DTI parameters (ADC, FA, and axial and radial diffusivity) are again summarized using crude means and
SDs, with differences between the two groups assessed using
mixed models with all regions included in a single model based
on a method used previously, both with and without adjusting
for BPD, gestational age at birth, and postnatal age at time of
MRI.16 Correlations among repeated observations within an
individual (multiple regions within the same infant) were modeled by including a random intercept for each infant. Heteroscedasticity between regions of the brain was modeled by allowing both the within- and between-subject variability to vary by
region. Regions were then compared combining the results
across hemisphere after checking for potential interactions
(between group and hemisphere, region and hemisphere, and a
three-way interaction between group, region, and hemisphere).
The majority of the statistical analysis was performed using
The perinatal characteristics of the 70 infants who were in
both the CAP and VIBeS studies were similar between the
33 randomly treated with caffeine and the 37 treated with
placebo (Table 1). After the treatment was started the
short-term benefit of caffeine in lowering the rate of BPD
(Table 1) was consistent with the effect observed overall in
the CAP study.1 Infants in both groups were scanned at a
mean age of 40.2 corrected weeks of gestation.
Qualitative MR for Cerebral Abnormality
Caffeine treatment appeared to have little effect on the
extent of qualitative WM and GM abnormality (median
[interquartile range]; WM abnormality score, caffeine 8
[7, 9], placebo 8 [6, 9], p ¼ 0.59; GM abnormality
score, caffeine 4 [3, 6], placebo 5 [3, 6], p ¼ 0.76), or
on the proportions with moderate to severe WM or GM
abnormalities (WM, caffeine 18% [6/33], placebo 17%
[6/37], p ¼ 0.87; GM, caffeine 27% [9/33], placebo
31% [11/37], p ¼ 0.76).
Regional and Total Cerebral Volumes
There were no substantial differences between the caffeine and placebo groups for total or tissue-specific segmented cerebral volumes, including hippocampal volumes (Table 2), or for parcellated brain volumes (Table
3; overall p-value for caffeine effect ¼ 0.67).
Diffusion Tensor Imaging (DTI)
In the subset of infants with DTI data (n ¼ 28) there was
clear evidence of an effect associated with caffeine exposure
on diffusion measures, with the ADC, radial, and axial diffusivity all being reduced in caffeine-exposed infants
(Tables 4, 5; Fig. 2). The impact of caffeine on these measures varied by region, with the greatest effects in the superior brain regions, particularly the superior central (SC)
and sensorimotor regions. There was, however, little effect
of caffeine on the FA measure (p ¼ 0.38), with both radial
and axial diffusion being reduced to a similar extent. In
addition, little effect was seen on any of the DTI measures
within the posterior limb of the internal capsule (inferior
central [IC] region). Although there were significant differences in DTI measures between hemispheres, the effect
of caffeine exposure did not vary by hemisphere, hence the
effect of caffeine was combined across the two hemispheres. There was little evidence of caffeine being
Volume 68, No. 5
Doyle et al: Caffeine and the Brain
TABLE 1: Characteristics of Infants According to Caffeine Treatment
Caffeine Group (N533) N (%)
Placebo Group (N537) N (%)
994 (168)
946 (181)
28 (2)
27 (2)
Birthweight <2 SD
2 (6%)
2 (5%)
Antenatal corticosteroids
33 (100%)
31 (86%)b
Cesarean delivery
24 (73%)
26 (70%)
Female sex
19 (58%)
15 (41%)
Singleton birth
18 (55%)
22 (59%)
Birth weight, g
Gestational age, weeks
Outcomes before discharge home
Bronchopulmonary dysplasiad
Patent ductus arteriosus
Necrotizing enterocolitis
7 (21%)
19 (51%)
9 (27%)
16 (43%)
1 (3%)
2 (5%)
Mean (SD).
Unknown whether antenatal corticosteroids were used in one child.
Chi-square P-values.
In oxygen at 36 weeks’ corrected age.
Treated with indomethacin or surgery.
detrimental to any of the diffusion measures, with no significant increase in ADC, radial, or axial diffusivity or
reduction in FA in the caffeine group in any region.
Adjustment for gestational age and postnatal age at
time of the MRI had little effect on the results (Table 3;
Fig. 2). Adjustment for BPD as a potential confounder/
mediator of the beneficial effect of caffeine had little
effect on the differences observed with caffeine on the
DTI measures, except that the marginally significant dif-
ferences in ADC and axial and radial diffusivity in the
inferior anterior region became nonsignificant (Fig. 2).
The major finding of this study is that caffeine treatment
appeared to have a substantial effect on cerebral WM diffusion measures in very preterm infants, with no obvious
effects on MR-defined cerebral injury or macrostructural
cerebral development, ie, qualitative WM and GM
TABLE 2: Segmented Brain Tissue and Hippocampal Volumes (cm3)
Caffeine Group
Placebo Group
Mean difference
(95% CI)
Total brain volume
457.3 (62.9)
453.6 (77.5)
3.7 (33.6, 41.0)
Total brain tissue
404.9 (55.9)
406.8 (60.4)
1.9 (32.2, 28.5)
Cortical gray matter
167.9 (45.8)
163.1 (42.2)
4.8 (17.8, 27.4)
Unmyelinated white matter
212.6 (28.0)
220.5 (28.4)
7.9 (22.5, 6.7)
Deep nuclear gray matter
12.8 (4.3)
13.0 (3.9)
0.2 (2.3, 1.9)
Myelinated white matter
11.7 (6.3)
10.3 (4.9)
1.4 (1.4, 4.3)
Cerebrospinal fluid
52.3 (26.3)
46.7 (27.9)
5.6 (8.5, 19.7)
Hippocampus, right
1.13 (0.19)
1.09 (0.17)
Hippocampus, left
1.11 (0.19)
1.07 (0.18)
0.04 (0.05, 0.12)
Data are mean (SD), unless otherwise specified. CI ¼ confidence interval.
P-values from t-test, unless otherwise specified.
CI and P-value from a mixed regression model.
November, 2010
of Neurology
TABLE 3: Absolute Parcellated Brain Volumes (cm3) According to Caffeine Treatment
Mean difference
(95% CI)1
Adjusted Mean
(95% CI)11
Orbitofrontal – left
25.3 (5.6)
25.8 (6.1)
0.58 (2.53, 1.37)
0.59 (2.56, 1.37)
Orbitofrontal – right
22.3 (4.9)
22.8 (5.6)
Dorsal prefrontal – left
4.9 (2.7)
5.6 (2.9)
0.77 (1.84, 0.30)
0.78 (1.88, 0.31)
Dorsal prefrontal – right
4.0 (2.3)
4.7 (2.4)
Premotor – left
26.2 (6.5)
26.2 (6.7)
0.20 (2.30, 1.90)
0.21 (2.32, 1.89)
Premotor – right
24.1 (6.1)
24.3 (6.0)
Subgenual – left
9.5 (2.6)
10.4 (3.5)
1.01 (2.01, 0.00)
1.02 (2.04, 0.00)
Subgenual – right
8.8 (2.8)
9.7 (3.6)
Sensorimotor – left
31.8 (6.0)
29.2 (6.5)
2.26 (0.07, 4.44)
2.24 (0.03, 4.45)
Sensorimotor – right
30.2 (5.3)
28.2 (5.8)
Midtemporal – left
15.2 (3.4)
14.3 (3.1)
0.82 (0.3, 2.01)
0.81 (0.41, 2.03)
Midtemporal – right
14.4 (3.2)
13.4 (3.4)
Parieto-occipital – right
59.4 (11.1)
59.3 (10.7)
1.09 (4.87, 2.70)
1.10 (4.90, 2.70)
Parieto-occipital – left
60.4 (10.7)
62.5 (12.9)
Inferior occipital and
cerebellum – right
33.6 (7.4)
34.6 (9.0)
1.11 (3.93, 1.72)
1.12 (3.95, 1.71)
Inferior occipital and
cerebellum – left
34.2 (8.6)
35.2 (9.6)
Data are mean (SD), unless otherwise specified. CI ¼ confidence interval.
Raw estimates; þEstimates from a single model including all regions, allowing for a region-by-side interaction, adjusted for total
brain volume and multiple observations within an individual. þþEstimates from a single model including all regions, allowing for
a region-by-side interaction, adjusted for total brain volume, gestational age at birth, gestational age at MRI, BPD, and multiple
observations within an individual.
abnormalities, or segmented, parcellated, or hippocampal
volumes. The reductions in ADC and both axial and radial diffusivity represent more mature cerebral WM organization in the caffeine-treated infants. This is an important finding, as we are aware of no other studies
reporting the effects of caffeine on cerebral injury or development in preterm infants evaluated by MRI. Adjustment for the rates of BPD had little effect on any statistical conclusions, suggesting that caffeine may be exerting
a direct neuroprotective effect, rather than a secondary
effect due to improvements in respiratory function, and
hence perhaps less hypoxia in the newborn period.
In this study we found that ADC and radial diffusivity were significantly lower in the caffeine-treated
group compared with placebo primarily in the superior
regions and the anterior inferior regions. Our infants
were imaged at a stage of brain development where there
is little myelination but an abundance of premyelinating
oligodendrocytes in the human cerebral white matter.
During this maturational phase the early differentiating
oligodendrocytes are ensheathing the cerebral WM
axons,17–19 with decreases in extracellular space and water
content leading to reductions in ADC and radial diffusivity.20–22 The ADC and radial diffusivity values in the
caffeine-treated infants were consistent with preservation
of WM maturation compared with the placebo group.
In addition, we also noted a reduction in axial diffusivity in the caffeine-treated infants compared with the
placebo-treated group. Data from animal studies have
shown that primary injury to axons, such as Wallerian
degeneration, is associated with a decrease in axial diffusivity, presumably because of disruption of the fiber tracts
along which water molecules can diffuse.23,24 However,
in the chronic phase the axial diffusivity is likely to
increase due to axonal loss. The infants in this study
were imaged at term-corrected age, many weeks after
birth, and the likely timing of any cerebral injury, and
hence a reduction in axial diffusivity in the caffeinetreated infants compared with controls may be consistent
with preservation of axons. The presence of a greater
Volume 68, No. 5
Doyle et al: Caffeine and the Brain
TABLE 4: Diffusion Tensor Imaging Data in Caffeine Group (n515) Compared with Placebo Group (n513)
Treatment group
Apparent Diffusion
(10-3 mm2/s)
Axial Diffusivity
(10-3 mm2/s)
Radial Diffusivity
(10-3 mm2/s)
Posterior – right
17.2 (3.9)
15.5 (4.2)
1.57 (0.15)
1.78 (0.15)
1.80 (0.15)
1.98 (0.15)
1.44 (0.15)
1.57 (0.17)
Posterior – left
15.8 (2.7)
14.6 (4.9)
1.54 (0.13)
1.75 (0.16)
1.79 (0.13)
1.90 (0.13)
1.41 (0.12)
1.57 (0.14)
Central – right
20.3 (8.2)
16.7 (6.1)
1.45 (0.17)
1.65 (0.24)
1.74 (0.16)
1.97 (0.27)
1.33 (0.22)
1.51 (0.26)
Central – left
20.2 (5.7)
16.3 (7.4)
1.41 (0.11)
1.66 (0.26)
1.73 (0.12)
1.95 (0.24)
1.31 (0.13)
1.51 (0.28)
Anterior – right
12.1 (2.7)
12.5 (4.0)
1.61 (0.14)
1.76 (0.16)
1.84 (0.17)
1.98 (0.15)
1.51 (0.14)
1.65 (0.17)
Anterior – left
15.0 (4.1)
13.4 (4.9)
1.56 (0.13)
1.71 (0.17)
1.79 (0.12)
1.95 (0.13)
1.44 (0.15)
1.57 (0.18)
Posterior – right
15.9 (3.3)
16.6 (4.8)
1.61 (0.12)
1.73 (0.12)
1.86 (0.12)
2.02 (0.11)
1.48 (0.13)
1.59 (0.14)
Posterior – left
16.0 (2.9)
16.8 (4.7)
1.68 (0.14)
1.69 (0.22)
1.96 (0.13)
1.95 (0.24)
1.55 (0.14)
1.54 (0.24)
Internal capsule –
48.6 (5.2)
51.5 (4.7)
1.09 (0.04)
1.08 (0.04)
1.75 (0.06)
1.77 (0.05)
0.76 (0.06)
0.73 (0.06)
Internal capsule –
46.0 (5.8)
49.2 (5.4)
1.09 (0.05)
1.10 (0.04)
1.74 (0.08)
1.79 (0.11)
0.79 (0.09)
0.77 (0.06)
Anterior – right
14.0 (4.3)
11.3 (7.4)
1.67 (0.14)
1.82 (0.20)
1.91 (0.12)
2.06 (0.17)
1.55 (0.17)
1.70 (0.22)
Anterior – left
13.6 (2.1)
13.4 (3.0)
1.65 (0.12)
1.72 (0.14)
1.89 (0.12)
1.97 (0.17)
1.54 (0.11)
1.63 (0.16)
Superior slice
Inferior slice
Data presented as crude mean (SD).
density of axons (ie, more closely packed) in the caffeinetreated infants in comparison with the placebo group
may also be a possible explanation for the finding of
reduced axial diffusivity with caffeine therapy.
There has been conflicting experimental evidence
from immature rodent models regarding the impact of
caffeine on cerebral injury and development. Caffeine
administration to postnatal mice (P3–P12) exposed to
hypoxia showed reduced ventriculomegaly and increased
myelination compared with placebo, suggestive of a
selective improvement in WM integrity.7 The neuroprotective effects of caffeine in this model were thought to
be mediated principally through adenosine receptor antagonism. The activation of the A1 adenosine receptor,
although neuroprotective in the mature brain,25 is deleterious in the immature brain.7,26 An earlier study also
showed that chronic low-dose caffeine administration in
an immature hypoxic-ischemic rat model was neuroprotective.6 In a rodent model without injury, there was
evidence of a potential beneficial effect of caffeine on
the immature brain, with chronic high-dose caffeine exposure over the first 12 days of life resulting in an
increase in dendritic length in layer-3 pyramidal neuNovember, 2010
rons within the prefrontal cortex.8 The authors speculated that the changes in dendritic arborization may be
associated with improved cognitive functioning in children and rats receiving caffeine.
In contrast, other experimental studies have found
potential neurotoxicity and behavioral dysfunction following caffeine administration. Studies have documented
alterations in astrocytogenesis,27 increased neuronal cell
death,9,28,29 reduction in proliferating glial cells,3 and
decreased weight and altered fatty acid composition in
the cerebellum in caffeine-treated rat and mice pups.30
There have also been reports of behavioral dysfunction
including hyperalgesia,4 hypoactivity,31 and impaired spatial learning tasks that persist into adulthood in rats
exposed to caffeine within the first week.4 Furthermore,
caffeine exposure of neonates could affect cerebral neurotransmission, as it has been reported to increase acetylcholinesterase activity in the hippocampus of adolescent
rodents.32 Chronic exposure of neonatal rodents to lowdose caffeine has also been demonstrated to reduce seizure
susceptibility33 and increase seizure threshold to some
convulsants,34 although the underlying mechanisms are
not certain.
of Neurology
TABLE 5: Summary of Fitting Mixed Models to Brain Microstructure as Determined by DTI (N 5 28)
Group (Caffeine vs not)
Region x Hemisphere
Region x Group
df ¼ degrees of freedom; F ¼ test statistic for null hypothesis that the factor effect is zero; P ¼ corresponding P-value.
Note these results are from a single model for each diffusion weighted imaging variable including all regions; there is no interaction between hemisphere and group and no 3-way interaction as these were not found to be significant in all four models.
Assesses overall effect of caffeine across regions.
Assesses whether the effect of caffeine varies by region.
FIGURE 2: Mean differences and 95% confidence intervals in diffusion-weighted imaging variables for children on caffeine
compared with those not on caffeine. Solid line represents the results from the unadjusted analyses and dashed line
represents results adjusted for gestational age at birth, gestational age at MRI and BPD. SP 5 superior posterior; SC 5
superior central; SA 5 superior anterior; IP 5 inferior posterior; IC 5 internal capsule; IA 5 inferior anterior. (A) Fractional
anisotropy (%). (B) Apparent diffusion coefficient (1023 mm2/S). (C) Radial diffusivity (1023 mm2/S). (D) Axial diffusivity (1023
Volume 68, No. 5
Doyle et al: Caffeine and the Brain
The changes in the MR measures observed in the
current study are not consistent with any evidence of
neurotoxicity from caffeine in humans of birthweight
<1,251 g when treated in the first weeks after birth.
There was little evidence of increased cerebral WM or
GM abnormality or any reductions in cerebral
volumes by term-equivalent age with chronic caffeine
A major strength of this study is that the data were
obtained from a randomized controlled trial of caffeine
therapy in humans, which is unlikely to be replicated
because of the clear benefits of caffeine.1,2 A further
strength of the study is that all MR data were obtained,
analyzed, and interpreted independent of knowledge of
caffeine treatment group. The effects of caffeine exposure
on the diffusion measures were large and are consistent
with previous animal experimental data suggestive of a
selective improvement in hemispheric WM integrity in
the immature brain, particularly in the setting of hypoxia. One may rationally question why such improvements in WM integrity were visible only with diffusion
measures with no related increase in WM volumes or
reduction in qualitative WM abnormality scores. However, the qualitative MR scoring is a subjective measure
defining injury that may be focal. In addition, the timing
of greatest vulnerability for WM injury may be in the
first 72 hours of life during physiological instability,
which occurred prior to the commencement of the caffeine therapy in most infants. It is plausible that earlier
administration of caffeine in the first 12 hours of life
may have greater impact on cerebral injury. The failure
to document any impact of caffeine therapy on WM volumes may relate to inherent errors in the methods, such
as reproducibility (WM 0.73–0.83) relating principally
to partial voluming errors. Thus, sensitivity for the
milder spectrum of alterations in WM integrity may
have been reduced compared with that of the diffusion
measures. It would be preferable to have data on more
subjects, and low power might have been an issue when
looking for differences with caffeine for the qualitative
and quantitative data. However, low power was not an
issue with respect to the DTI data, where clear differences were detected on even fewer infants.
A potential weakness of the study relates to the diffusion methods involved in evaluation of the cerebral
WM. The resolution during diffusion acquisition, the
limited number of directions and b-values, and the
method of analysis with manual regions of interest were
the available techniques at the time of study. Clearly, diffusion MR methods have advanced substantially since our
images were acquired. More recent acquisition protocols
and processing of diffusion images would likely enable a
November, 2010
more comprehensive evaluation of WM integrity, particularly in relation to WM tracts in multiple cerebral regions
without regional preselection, such as tract-based spatial
statistics (TBSS). However, ROI-based analysis methods
are still frequently employed in the evaluation of the preterm infant,35,36 with meaningful insights. In addition,
line scan diffusion image acquisition and analysis was
chosen for these infants due to improved signal-to-noise,
reduced impact of motion and eddy current related artifacts, and reduced intersubject variability.13 A final limitation relates to the issue of multiple statistical tests. We
consider each area of testing to be addressing individual
hypotheses; however, we reduced the chance of type I
error by fitting single mixed models for the DTI data
combining the data across hemisphere and region.
In summary, chronic caffeine treatment in preterm
infants appears to improve cerebral WM development in
the human brain, which may explain how caffeine
improved neurodevelopment in the immature human. As
the children in both the CAP and the VIBeS studies are
being reassessed at 7–8 years of age, a repeat MRI at that
time will help to determine the long-term impact of neonatal caffeine therapy on brain structure in early childhood.
This work was supported by the National Health and
Medical Research Council of Australia (Project Grants
No. 237117 and No. 108706), the Royal Women’s Hospital Research Foundation, and the Brockhoff Foundation.
Role of the funding sources: The funding sources
had no role in the design, implementation, analysis,
interpretation, or writing of the results of this study, or
in the decision to submit.
We thank the CAP Steering Committee, chaired by
Barbara Schmidt, for permitting the CAP study statistician to release the CAP study randomization sequence
for the children who are included in this report, but only
to the statistician in Melbourne who performed the present analyses.
Potential Conflicts of Interest
Nothing to report.
Schmidt B, Roberts RS, Davis P, et al. Caffeine therapy for apnea
of prematurity. N Engl J Med 2006;354:2112–2121.
Schmidt B, Roberts RS, Davis P, et al. Long-term effects of caffeine
therapy for apnea of prematurity. N Engl J Med 2007;357:1893–1902.
Marret S, Delpech B, Girard N, et al. Caffeine decreases glial cell
number and increases hyaluronan secretion in newborn rat brain
cultures. Pediatr Res 1993;34:716–719.
of Neurology
Pan HZ, Chen HH. Hyperalgesia, low-anxiety, and impairment of
avoidance learning in neonatal caffeine-treated rats. Psychopharmacology (Berl) 2007;191:119–125.
Wimberger DM, Roberts TP, Barkovich AJ, et al. Identification of
‘‘premyelination’’ by diffusion-weighted MRI. J Comput Assist
Tomogr 1995;19:28–33.
Sutherland GR, Peeling J, Lesiuk HJ, et al. The effects of caffeine
on ischemic neuronal injury as determined by magnetic resonance
imaging and histopathology. Neuroscience 1991;42:171–182.
Bona E, Aden U, Fredholm BB, Hagberg H. The effect of long
term caffeine treatment on hypoxic-ischemic brain damage in the
neonate. Pediatr Res 1995;38:312–318.
Huppi PS, Maier SE, Peled S, et al. Microstructural development
of human newborn cerebral white matter assessed in vivo by diffusion tensor magnetic resonance imaging. Pediatr Res 1998;44:
Miller SP, Vigneron DB, Henry RG, et al. Serial quantitative
diffusion tensor MRI of the premature brain: development in newborns with and without injury. J Magn Reson Imaging 2002;16:
Pierpaoli C, Barnett A, Pajevic S, et al. Water diffusion changes in
Wallerian degeneration and their dependence on white matter
architecture. Neuroimage 2001;13:1174–1185.
Song SK, Sun SW, Ju WK, et al. Diffusion tensor imaging
detects and differentiates axon and myelin degeneration in
mouse optic nerve after retinal ischemia. Neuroimage 2003;20:
Woodward LJ, Anderson PJ, Austin NC, et al. Neonatal MRI to
predict neurodevelopmental outcomes in preterm infants. N Engl
J Med 2006;355:685–694.
Fredholm BB, Battig K, Holmen J, et al. Actions of caffeine in the
brain with special reference to factors that contribute to its widespread use. Pharmacol Rev 1999;51:83–133.
Thompson DK, Wood SJ, Doyle LW, et al. Neonate hippocampal
volumes: Prematurity, perinatal predictors, and 2-year outcome.
Ann Neurol 2008;63:642–651.
Rivkees SA, Zhao Z, Porter G, Turner C. Influences of adenosine on the fetus and newborn. Mol Genet Metab 2001;74:
Thompson DK, Wood SJ, Doyle LW, et al. MR-determined hippocampal asymmetry in full-term and preterm neonates. Hippocampus 2009;19:118–123.
Desfrere L, Olivier P, Schwendimann L, et al. Transient inhibition
of astrocytogenesis in developing mouse brain following postnatal
caffeine exposure. Pediatr Res 2007;62:604–609.
Kubicki M, Maier SE, Westin CF, et al. Comparison of single-shot
echo-planar and line scan protocols for diffusion tensor imaging.
Acad Radiol 2004;11:224–232.
Kang SH, Lee YA, Won SJ, et al. Caffeine-induced neuronal death
in neonatal rat brain and cortical cell cultures. Neuroreport 2002;
Stjeskal E, Tanner J. Spin diffusion measurements: spin echoes in
the presence of a time dependent field gradient. J Chem Phys
Black AM, Pandya S, Clark D, et al. Effect of caffeine and morphine on the developing pre-mature brain. Brain Res 2008;1219:
Basser PJ, Pierpaoli C. Microstructural and physiological features
of tissues elucidated by quantitative-diffusion-tensor MRI. J Magn
Reson B 1996;111:209–219.
Yazdani M, Ide K, Asadifar M, et al. Effects of caffeine on the saturated and monounsaturated fatty acids of the newborn rat cerebellum. Ann Nutr Metab 2004;48:79–83.
Thompson DK, Warfield SK, Carlin JB, et al. Perinatal risk factors
altering regional brain structure in the preterm infant. Brain 2007;
Zimmerberg B, Carr KL, Scott A, et al. The effects of postnatal caffeine exposure on growth, activity and learning in rats. Pharmacol
Biochem Behav 1991;39:883–888.
da Silva RS, Richetti SK, da Silveira VG, et al. Maternal caffeine
intake affects acetylcholinesterase in hippocampus of neonate
rats. Int J Dev Neurosci 2008;26:339–343.
Tchekalarova J, Kubova H, Mares P. Effects of postnatal caffeine
exposure on seizure susceptibility in developing rats. Brain Res
Guillet R, Dunham L. Neonatal caffeine exposure and seizure susceptibility in adult rats. Epilepsia 1995;36:743–749.
Skiold B, Horsch S, Hallberg B, et al. White matter changes in
extremely preterm infants, a population-based diffusion tensor
imaging study. Acta Paediatr 2010;99:842–849.
Hart AR, Whitby EH, Clark SJ, et al. Diffusion-weighted imaging
of cerebral white matter and the cerebellum following preterm
birth. Dev Med Child Neurol 2010;52:652–659.
Back SA, Craig A, Luo NL, et al. Protective effects of caffeine on
chronic hypoxia-induced perinatal white matter injury. Ann Neurol
Juarez-Mendez S, Carretero R, Martinez-Tellez R, et al. Neonatal
caffeine administration causes a permanent increase in the dendritic length of prefrontal cortical neurons of rats. Synapse 2006;60:
Alvira D, Yeste-Velasco M, Folch J, et al. Neuroprotective effects
of caffeine against complex I inhibition-induced apoptosis are
mediated by inhibition of the Atm/p53/E2F-1 path in cerebellar
granule neurons. J Neurosci Res 2007;85:3079–3088.
Kinney HC, Brody BA, Kloman AS, Gilles FH. Sequence of central
nervous system myelination in human infancy. II. Patterns of myelination in autopsied infants. J Neuropathol Exp Neurol 1988;47:
Back SA, Luo NL, Borenstein NS, et al. Late oligodendrocyte progenitors coincide with the developmental window of vulnerability
for human perinatal white matter injury. J Neurosci 2001;21:
Back SA, Luo NL, Borenstein NS, et al. Arrested oligodendrocyte
lineage progression during human cerebral white matter development: dissociation between the timing of progenitor differentiation
and myelinogenesis. J Neuropathol Exp Neurol 2002;61:197–211.
Volume 68, No. 5
Без категории
Размер файла
245 Кб
development, infant, caffeine, preterm, brain
Пожаловаться на содержимое документа