ADHD is a Developmental Disorder

Please read this longitudinal study completed by Shaw et al. in 2007 at the National Institute of Health who proved definitively that ADHD is a developmental disorder. I will comment on this article in my next post.

Attention-deficit/hyperactivity disorder is
characterized by a delay in cortical maturation
P. Shaw†‡, K. Eckstrand†, W. Sharp†, J. Blumenthal†, J. P. Lerch§, D. Greenstein†, L. Clasen†, A. Evans§,
J. Giedd†, and J. L. Rapoport†
†Child Psychiatry Branch, National Institute of Mental Health, Room 3N202, Building 10, Center Drive, Bethesda, MD 20892; and §Montreal Neurological
Institute, McGill University, Montreal, QC, Canada H3A 2T5
Edited by Leslie G. Ungerleider, National Institutes of Health, Bethesda, MD, and approved October 5, 2007 (received for review August 17, 2007)
There is controversy over the nature of the disturbance in brain
development that underpins attention-deficit/hyperactivity disorder
(ADHD). In particular, it is unclear whether the disorder results
from a delay in brain maturation or whether it represents a
complete deviation from the template of typical development.
Using computational neuroanatomic techniques, we estimated
cortical thickness at >40,000 cerebral points from 824 magnetic
resonance scans acquired prospectively on 223 children with ADHD
and 223 typically developing controls. With this sample size, we
could define the growth trajectory of each cortical point, delineating
a phase of childhood increase followed by adolescent decrease
in cortical thickness (a quadratic growth model). From these trajectories,
the age of attaining peak cortical thickness was derived
and used as an index of cortical maturation. We found maturation
to progress in a similar manner regionally in both children with and
without ADHD, with primary sensory areas attaining peak cortical
thickness before polymodal, high-order association areas. However,
there was a marked delay in ADHD in attaining peak thickness
throughout most of the cerebrum: the median age by which 50%
of the cortical points attained peak thickness for this group was
10.5 years (SE 0.01), which was significantly later than the median
age of 7.5 years (SE 0.02) for typically developing controls (log rank
test (1)2 5,609, P < 1.0 10 20). The delay was most prominent
in prefrontal regions important for control of cognitive processes
including attention and motor planning. Neuroanatomic documentation
of a delay in regional cortical maturation in ADHD has not
been previously reported.
cortical development structural neuroimaging
Attention-deficit/hyperactivity disorder (ADHD) is the most
common neurodevelopment disorder of childhood affecting
between 3% and 5% of school-aged children (1). Since its
earliest descriptions, there has been debate as to whether the
disorder is a consequence partly of delay in brain maturation or
as a complete deviation from the template of typical development
(2). Several studies find that brain activity at rest and in
response to cognitive probes is similar between children with
ADHD and their slightly younger but typically developing peers,
evidence congruent with a maturational lag in cortical development
(3–5). However, others report a quantitatively distinct
neurophysiology, with a unique architecture of the electroencephalogram
and some highly anomalous findings in functional
imaging studies, more in keeping with ADHD as a deviation
from typical development (6–10).
In a previous longitudinal study, we found parallel trajectories
of gray lobar volume change in children with ADHD and
typically developing controls, but more focal changes in cortical
maturation occurring at a sublobar level would not be detected
by this lobar measure (11). We thus aimed to define the
trajectory of cortical development using a measure of cortical
thickness that affords exquisite spatial resolution. Cortical thickness
was chosen as a metric that both captures the columnar
architecture of the cortex and is sensitive to developmental
change in typically developing and clinical populations (12–15).
Most of the 446 children in the current study had repeated
neuroanatomic imaging—112 (25%) had two scans, 88 (20%)
had three scans, and 30 (7%) had four or more scans, performed
at a mean interval between scans of 2.8 years. Such longitudinal
data can be combined with cross-sectional data by using mixedmodel
regression to model developmental change, with the
longitudinal data being particularly informative. For cortical
thickness data, the simplest trajectory that can be fitted to
describe its change over time is a straight line. More complex
growth models include distinct phases of increase and decrease
in cortical thickness: A quadratic model has two such phases
(typically an initial increase that reaches a peak before declining)
and a cubic model has three. Derived properties of these
developmental curves are frequently used as developmental
indices, such as the age at which points of inflection in the curve
are attained (16, 17). When considering cortical change, the age
at which peak cortical thickness is reached—the point where
increase gives way to decrease in cortical thickness—emerges as
a particularly useful index. Note that the ability to detect a
quadratic or cubic growth model is a prerequisite for defining the
age of peak cortical thickness; it cannot be determined from a
linear model.
We thus compared the age of attaining peak cortical thickness in
children with and without ADHD to determine whether the
disorder is characterized by a delay in cerebral cortical maturation.
Results
The temporal sequence of cortical maturation, reflected by the
age of reaching peak cortical thickness at cortical points where
a quadratic model was appropriate, was similar in both groups
[see supporting information (SI) Movies 1 and 2 and Fig. 1]. In
the frontal cortex, the superior, precentral, and polar regions
reached an early peak in cortical thickness, followed by a
centripetal wave moving toward the middle prefrontal cortex. In
the temporal cortex, posterior portions of the middle and
superior temporal cortex matured before more anterior temporal
regions. In the occipital cortex, for both the typically developing
and ADHD subjects, there were early peaks with little
developmental change in the age period covered. Direct comparison
of cortical change in the parietal regions was complicated
because the groups differed in the regions where a
quadratic model was appropriate.
However, although the overall pattern of development was
similar, there were pronounced diagnostic differences in timing.
Where a peak age could be determined, the ADHD group
Author contributions: P.S. and J.L.R. designed research; P.S., W.S., J.B., L.C., J.G., and J.L.R.
performed research; K.E., J.B., J.P..L., and A.E. contributed new reagents/analytic tools; P.S.,
K.E., and D.G. analyzed data; and P.S. and J.L.R. wrote the paper.
The authors declare no conflict of interest.
This article is a PNAS Direct Submission.
‡To whom correspondence should be addressed. E-mail: shawp@mail.nih.gov.
This article contains supporting information online at www.pnas.org/cgi/content/full/
0707741104/DC1.
© 2007 by The National Academy of Sciences of the USA
www.pnas.org cgi doi 10.1073 pnas.0707741104 PNAS December 4, 2007 vol. 104 no. 49 19649–19654
PSYCHOLOGY
generally reached this milestone later than the typically developing
controls; see Fig. 2. Kaplan–Meier curves showed that the
median age by which 50% of the cortical points had attained
peak thickness for the ADHD group was 10.5 years (SE 0.01),
which was significantly later than the median age of 7.5 years (SE
0.02) for the typically developing controls (log-rank test (1)2
5,609, P 1.0 10 20); Fig. 3. Differences were most prominent
in the middle prefrontal cortex, where theADHDgroup reached
their peak thickness 5 years after the typically developing
controls, and to a lesser extent in the superior prefrontal and
medial prefrontal cortex (with the ADHD group peaking 2
years later). Kaplan–Meier curves for the prefrontal region
demonstrated that, although both groups had a similar rates of
attaining cortical thickness, this was delayed in theADHDgroup
with a median age of 10.4 years (SE 0.02), compared with
typically developing control median age of 7.5 years (SE 0.02)
(log-rank test (1)2 9,599, P 1.0 10 20). Posteriorly, delay
was present bilaterally in the middle and superior temporal
cortex, extending to the middle occipital gryi, with the ADHD
group having a peak age of 10.6 years (SE 0.04) and the
typically developing controls peaking at 6.8 years (SE 0.08)
log-rank test (1)2 303, P 1.0 10 20).
The ADHD group had an earlier peak thickness predominately
in the primary motor cortex, with a median age by which
50% of points within this region peaked at 7 years (SE 0.16)
compared with 7.4 years (SE 0.12) for the typically developing
controls (log-rank test (1)2 10, P 0.001); Fig. 4.
The pattern of results held when the degree of motion artifact
was entered into the regression equation (see SI Figs. 5 and 6).
Discussion
Cortical development in children with ADHD lagged behind
that of typically developing children by several years. However,
the ordered sequence of regional development, with primary
sensory and motor areas attaining their peak cortical thickness
before high-order association areas, was similar in both groups,
suggesting that ADHD is characterized by delay rather than
deviance in cortical maturation. This contrasts with other neurodevelopmental
disorders such as autism in which there appears
to be a dramatic shift of brain growth curves to the right along
the age axis, resulting in peak brain volumes being reached at a
7 8 9 10 11 12
ADHD
Typically developing controls
7 8 9 10
ADHD
Typically developing controls
11 12 13
A
B
Fig. 1. The age of attaining peak cortical thickness in children with ADHD compared with typically developing children. (A) dorsal view of the cortical regions
where peak thickness was attained at each age (shown, ages 7–12) in ADHD (Upper) and typically developing controls (Lower). The darker colors indicate regions
where a quadratic model was not appropriate (and thus a peak age could not be calculated), or the peak age was estimated to lie outside the age range covered.
Both groups showed a similar sequence of the regions that attained peak thickness, but the ADHD group showed considerable delay in reaching this
developmental marker. (B) Right lateral view of the cortical regions where peak thickness was attained at each age (shown, ages 7–13) in ADHD (Upper) and
typically developing controls (Lower). Again, the delay in ADHD group in attaining peak cortical thickness is apparent.
Greater than 2 yearsʼ delay
0 to 2 years delay
Fig. 2. Regions where the ADHD group had delayed cortical maturation, as
indicated by an older age of attaining peak cortical thickness.
19650 www.pnas.org cgi doi 10.1073 pnas.0707741104 Shaw et al.
much earlier age—the opposite of the pattern we note in ADHD
(18, 19).
The cortical maturation delay in ADHD was most prominent
in the lateral prefrontal cortex, the region with the most
consistent reports of structural anomalies in the disorder (11,
20), particularly within the superior and dorsolateral prefrontal
regions (21–23). The prefrontal cortex supports a host of cognitive
functions, such as the ability to suppress inappropriate
responses and thoughts (24, 25), the executive ‘‘control’’ of
attention (26), evaluation of reward contingencies (27, 28),
higher-order motor control (5), and working memory (29).
Deficits in these cognitive functions have all been implicated in
the pathogenesis of ADHD (30), and prefrontal cortical hypoactivation
in children with ADHD during performance of many
of these tasks is a relatively consistent finding (10).
Delay was also found in the temporal cortex, most prominently
in the posterior portions of the middle/superior temporal gyrus
bilaterally, relatively circumscribed on the left, and with more
posterior extension on the right. Structural change in the temporal
lobes is a common finding in studies of ADHD, from the
level of the entire lobe (11) through more focal gray matter
density and cortical thickness anomalies (31, 32) and may have
metabolic (9, 33), functional (10, 34–36), and electrophysiological
correlates (37, 38). A unifying feature of the frontal and
temporal regions with greatest maturational delay is the involvement
of heteromodal cortex (39). These are interconnected
cortical regions that integrate information from lower-order
sensory areas giving higher-order percepts that guide the control
of attention and action. Structural anomalies of this system have
been implicated in the pathogenesis of ADHD (31).
By contrast, the primary motor cortex was the only cortical
area in which the ADHD group showed slightly earlier maturation.
It is possible that the combination of early maturation of
the primary motor cortex with late maturation of higher-order
motor control regions may reflect or even drive the excessive and
poorly controlled motor activity cardinal to the syndrome.
Reaching peak cortical thickness at a younger age also means
the typically developing children enter earlier the phase of
cortical thinning that dominates adolescence (40, 41). Because of
the limited age range, we were not able to define the age at which
the adolescent phase of cortical thinning levels off, transitioning
into stable adult cortical dimensions. We predict that the age of
reaching this essentially static adult phase would also be later in
the subjects with ADHD.
To our knowledge, neuroanatomic evidence supportive of the
theory of delay in cortical maturation in ADHD has not been
previously reported. The use of a cortical measure that affords
exquisite spatiotemporal resolution allows us to demonstrate
considerable variability in timing of cortical maturation within
each lobe not detectable by our earlier lobar volumetric analyses
(11). Additionally, we are able to localize the greatest maturational
delay to prefrontal cortical regions implicated in the
pathogenesis of ADHD.
In other work on a subsample of subjects with clinical outcome
data from this cohort, we were only able to detect linear patterns
of change in cortical thickness (and thus could not define the age
of peak cortical thickness) and found generally parallel trajectories
with the exception of a region in the right parietal cortex
(12). By including additional subjects, we are able to detect
higher-order effects of age and thus map out diagnostic regional
differences in the age of attaining peak cortical thickness (12).
Because we lacked clinical outcome data on the majority of the
ADHD subjects in the current study, we were unable to examine
the possibility that good or poor clinical outcome is linked to
differences in the timing of key developmental markers, such as
the age of peak cortical thickness.
Returning to the central finding, the generally older age of
attaining peak cortical thickness in ADHD presumably represents
a temporal shift in the balance between the cellular
Fig. 3. Kaplan–Meier curves illustrating the proportion of cortical points that had attained peak thickness at each age for all cerebral cortical points (Left) and the
prefrontal cortex (Right). The median age by which 50% of cortical points had attained their peak differed significantly between the groups (all P 1.0 10 20)
Fig. 4. Regions where the ADHD group had early cortical maturation, as
indicated by a younger age of attaining peak cortical thickness.
Shaw et al. PNAS December 4, 2007 vol. 104 no. 49 19651
PSYCHOLOGY
processes that result in an initial increase and later decrease in
cortical thickness. The exact nature of these processes in typically
developing children is yet to be determined. Extrapolating from
animal studies, the increase in cortical thickness may be driven
by mechanisms such as dendritic spine growth and elaboration
of supporting glia and vasculature (42, 43). Cortical thinning in
adolescence may reflect intracortical myelination and the usedependent
selective elimination of synapses that may help create
and sculpt neural circuits, including those supporting cognitive
abilities (44–46). Turning to ADHD, animal models are mostly
based on perturbations in monoaminergic neurotransmission
arising in response to either early insults (e.g., induced transient
hyperthyroidism, or neonatal 6-OHDA lesions) or anomalies of
neurotransmitters (such as the 160-bp insertion in exon 3 of the
dopamine transport gene in the spontaneously hypertensive rat)
(47, 48). How such changes might influence the dynamics of
cortical development remains unclear but would be an important
area for future research.
What etiological factors might underpin this delay? Trophic
effects of treatment with psychostimulants in the ADHD group
are possible but unlikely, given our previous reports of no effect
of psychostimulants on gray matter volume (11). Because our
studies have been observational, however, any conclusions about
stimulants are tentative. Our overall results cannot be attributed
to group differences in intelligence and gender, which, although
they effect cortical development (14, 41, 49, 50), were strictly
controlled in our design. Genetic factors will certainly play a role,
with a perturbation in the developmental sequence of the
activation and deactivation of genes that sculpt cortical architecture.
In this context, neurotrophins, essential for the proliferation,
differentiation, and survival of neuronal and nonneuronal
cells, emerge as promising candidates, and, indeed,
polymorphisms within the brain-derived neurotrophic factor and
nerve growth-factor 3 genes have already been tentatively linked
with ADHD (51, 52).
Trajectories of brain development built on longitudinal and
cross-sectional neuroanatomic data sets are providing rich insights
into ADHD. Not only do they informkey debates that have
existed since the earliest descriptions of the disorder (2), but they
may also guide the future search for factors that delay, rather
than derail, cortical development.
Methods
Subjects. The clinical group comprised 223 children and adolescents
with Diagnostic and Statistical Manual of Mental Disorders,
Fourth Edition (DSM-IV)-defined ADHD. Diagnosis was
based on the Parent Diagnostic Interview for Children and
Adolescents (53), Conner’s Teacher Rating Scales (54), and the
Teacher Report Form; see Table 1. Exclusion criteria were IQ
under 80 and evidence of medical or neurological disorders. Two
hundred five (92%) had combined-type ADHD at baseline, 13
(6%) had inattentive subtype, and 5 (2%) had hyperactive/
impulsive subtype. One hundred fifty-four unrelated singletons
and 25 sets of affected singleton siblings (with 53 individuals)
and 16 twin-births (only one child per twin-pair) were included.
Typically developing controls were recruited, and each subject
completed the Childhood Behavior Checklist as a screening tool
and then underwent a structured diagnostic interview by a child
psychiatrist to rule out any psychiatric or neurological diagnoses
(55). The typically developing participants in this study were
matched to the ADHD group on gender, age, and intelligence as
measured by age-appropriate version of the Wechsler Intelligence
Scales. There were 169 singletons, 17 sets of unaffected
siblings (with 38 individuals), and 16 twin births (one child per
twin pair). The institutional review board of the National
Institutes of Health approved the research protocol, and written
informed consent and assent to participate in the study were
obtained from parents and children, respectively.
The total number of subjects scanned at each age is given in
Table 2, which also shows the numbers of subjects undergoing
repeated scanning and the mean age at each wave of scan
acquisition. The mean interscan interval was 2.9 years (SD 1.5)
for the ADHD group and 2.8 years (SD 1.4) for the typically
developing controls [t (317) 1.3, P 0.2].
Neuroimaging. All children had neuroanatomic magnetic resonance
imaging on the same 1.5-T General Electric Signa scanner throughout
the study. Imaging parameters were echo time of 5 ms,
repetition time of 24 ms, flip angle of 45°, acquisition matrix of
256 192, number of excitations equaling 1, and 24-cm field of
view. Head placement was standardized as described (56). The
same 1.5-T General Electric Signa scanner was used throughout the
study. The native MRI scans were registered into standardized
stereotaxic space by using a linear transformation and corrected for
Table 1. Demographic and clinical details of the subjects
Characteristic
ADHD, N
223
Controls, N
223
Age at initial scan, yr, mean (SD)* 10.2 (3.2) 10.6 (3.5)
Gender
Male:female 141:82 141:82
Estimated IQ,† mean (SD) 109 (15) 111 (13)
Comorbid diagnoses
Oppositional defiant disorder, no. (%) 77 (35) NA
Conduct disorder, no. (%) 15 (7) NA
Learning disorder, no. (%) 19 (9) NA
Mood disorder, no. (%) 8 (4) NA
Anxiety disorder, no. (%) 13 (6) NA
Tic (NOS), no. (%) 14 (6) NA
Clinical details
Clinical Global Assessment Scale, mean (SD) 48 (7) NA
CBCL Attention Problems T score, mean (SD) 71 (8) NA
TRF Attention Problems T score, mean (SD) 66 (10) NA
Prior stimulant treatment, no. (%) 108 (66) NA
NA, not applicable; NOS, not otherwise specified; CBCL, Child Behavior Checklist; TRF, Teacher Report Form.
*P (ADHD vs controls), t (444) 1.4, P 0.16.
†P (ADHD vs controls), t (426) 1.5, P 0.14
19652 www.pnas.org cgi doi 10.1073 pnas.0707741104 Shaw et al.
nonuniformity artifacts (57). The registered and corrected volumes
were segmented into white matter, gray matter, cerebrospinal fluid,
and background by using an advanced neural net classifier (58). The
inner and outer cortical surfaces were then extracted by using
deformable models and nonlinearly aligned toward a standard
template surface (59). Cortical thickness was then measured in
native space millimeters by using the linked distance between the
pial white and gray matter surfaces at 40,960 vertices throughout the
cerebral cortex. In estimating cortical thickness, we chose a 30-
mm-bandwidth blurring kernel on the basis of a population simulation
study, which showed that this bandwidth maximized statistical
power while minimizing false positives (60). This kernel
preserves the capacity for anatomical localization because 30-mm
blurring along the surface by using a diffusion smoothing operator
preserves cortical topological features and represents considerably
less cortex than the equivalent volumetric Gaussian blurring kernel
(60). All scans were rated for degree of motion artifact (none, mild,
moderate, or severe), as detailed in SI Text and ref. 61. Scans with
moderate or severe motion artifact were excluded from further
analyses; scans with mild motion artifact were included.
Statistical Analyses. First, we determined developmental trajectories,
using mixed model regression analysis that allows the
inclusion of multiple measurements per person, missing data,
and irregular intervals between measurements, thereby increasing
statistical power (62). A random effect for each individual
was nested within a random effect for each family, thus accounting
for both within-person and within-family dependence. Our
classification of developmental trajectories was based on a
step-down model selection procedure: At each cortical point, we
modeled cortical thickness by using a mixed-effects polynomial
regression model, testing for cubic, quadratic, and linear age
effects. If the cubic age effect was not significant at P 0.05, it
was removed, and we stepped down to the quadratic model and
so on. In this way, we were able to classify the development of
each cortical point as being best explained by a cubic, quadratic,
or linear function of age. A quadratic model proved appropriate
for much of the cortex, in which kth cortical thickness of the ith
individual in the jth family was modeled as Thicknessijk
intercept dij 1(age mean age) 2*(age mean
age)**2) eijk, where dij are nested random effects modeling
within-person and within family dependence, the intercept and
terms are fixed effects, and eijk represents the residual error.
Specifically, for both the ADHD and typically developing controls,
a quadratic model was appropriate throughout most of the
lateral prefrontal and medial prefrontal cortex, the superior and
middle temporal cortex, superior and middle occipital cortex,
and angular and supramarginal gyri. The ADHD group showed
a linear fit in the superior parietal lobules and postcentral gyri,
unlike the typically developing controls, for whom a quadratic
model held. The analyses were repeated, entering the degree of
motion artifact into the regression equation.
Next, the age of reaching peak cortical thickness for each
group was calculated in these regions from the first-order
derivatives of the fitted curves and illustrated through dynamic
time-lapse sequences (‘‘movies’’). Kaplan–Meier curves were
constructed showing the proportion of cortical points that had
reached peak cortical thickness throughout the age range covered.
The significance of the group difference in the median age
by which half of the cortical points had attained their peak
thickness was calculated by using the log-rank (Mantel–Cox) test.
Brain maps show the regions where the ADHD group attained
peak thickness at either an earlier or later age.
We thank F. X. Castellanos for initiating the study and for advice and
support and the children and their families who participated in the study.
This work was supported by the Intramural Research Program of the
National Institutes of Health. The sponsor of the study had no role in
study design, data interpretation, or writing of the report.
1. Buitelaar JK (2002) in Hyperactivity and Attention-Deficit Disorders, ed Sandberg
J (Cambridge Univ Press, Cambridge, UK).
2. Kinsbourne M (1973) Ann NY Acad Sci 205:268–273.
3. Mann CA, Lubar JF, Zimmerman AW, Miller CA, Muenchen RA (1992)
Pediatr Neurol 8:30–36.
4. El-Sayed E, Larsson JO, Persson HE, Santosh PJ, Rydelius PA (2003) Acta
Paediatr 92:776–784.
5. Rubia K, Overmeyer S, Taylor E, Brammer M, Williams SC, Simmons A,
Bullmore ET (1999) Am J Psychiatry 156:891–896.
6. Clarke AR, Barry RJ, McCarthy R, Selikowitz M (2001) Psychophysiology
38:212–221.
7. Chabot RJ, Serfontein G (1996) Biol Psychiatry 40:951–963.
8. Hobbs MJ, Clarke AR, Barry RJ, McCarthy R, Selikowitz M (2007) Clin
Neurophysiol 118:363–371.
9. Zametkin AJ, Liebenauer LL, Fitzgerald GA, King AC, Minkunas DV,
Herscovitch P, Yamada EM, Cohen RM (1993) Arch Gen Psychiatry 50:333–
340.
10. Dickstein SG, Bannon K, Castellanos FX, Milham MP (2006) J Child Psychol
Psychiatry 47:1051–1062.
11. Castellanos FX, Lee PP, Sharp W, Jeffries NO, Greenstein DK, Clasen LS,
Blumenthal JD, James RS, Ebens CL, Walter JM, et al. (2002) J Am Med Assoc
288:1740–1748.
12. Shaw P, Lerch J, Greenstein D, Sharp W, Clasen L, Evans A, Giedd J,
Castellanos FX, Rapoport J (2006) Arch Gen Psychiatry 63:540–549.
13. Makris N, Biederman J, Valera EM, Bush G, Kaiser J, Kennedy DN, Caviness
VS, Faraone SV, Seidman LJ (2006) Cereb Cortex 17:1364–1375.
14. Shaw P, Greenstein D, Lerch J, Clasen L, Lenroot R, Gogtay N, Evans A,
Rapoport J, Giedd J (2006) Nature 440:676–679.
Table 2. Scan acquisition details
ADHD
Typically
developing controls
Number of subjects with scans at each age, yr*
5 and under 13 24
6 18 24
7 37 28
8 44 24
9 45 31
10 40 39
11 33 38
12 33 38
13 24 33
14 29 30
15 22 30
16 17 25
17 19 15
18 13 17
19 8 7
20 and over 13 13
No. of subjects at each wave of scan acquisition†
Time 1 223 223
Time 2 111 119
Time 3 59 59
Time 4 15 15
Mean age at each scan, yr (SD)
Time 1‡ 10.2 (3.2) 10.6 (3.5)
Time 2§ 12.8 (3.6) 13.3 (4.0)
Time 3¶ 15.3 (3.7) 14.8 (3.6)
Time 4 17.9 (3.6) 16.1 (3.3)
*Absolute number of scans obtained at each age for each group.
†Number of subjects in each group who had scans at each wave of acquisition
and the mean age (and SD) of each wave.
‡t (444) 1.4, P 0.16.
§t (228) 1.1, P 0.26.
¶t (116) 0.73, P 0.46.
t (28) 1.46, P 0.15.
Shaw et al. PNAS December 4, 2007 vol. 104 no. 49 19653
PSYCHOLOGY
15. Lerch J, Pruessner JC, Zijdenbos A, Hampel H, Teipel SJ, Evans A (2005)
Cereb Cortex 15:995–1001.
16. Tanner JM, Whitehouse RH, Marubini E, Resele LF (1976) Ann Hum Biol
3:109–126.
17. Jolicoeur P, Pontier J, Pernin MO, Sempe M (1988) Biometrics 44:995–1003.
18. Courchesne E, Karns CM, Davis HR, Ziccardi R, Carper RA, Tigue ZD,
Chisum HJ, Moses P, Pierce K, Lord C, et al. (2001) Neurology 57:245–254.
19. Courchesne E, Carper R, Akshoomoff N (2003) J Am Med Assoc 290:337–344.
20. Durston S, Hulshoff Pol HE, Schnack HG, Buitelaar JK, Steenhuis MP,
Minderaa RB, Kahn RS, van Engeland H (2004) J Am Acad Child Adolesc
Psychiatry 43:332–340.
21. Filipek PA, Semrud-Clikeman M, Steingard RJ, Renshaw PF, Kennedy DN,
Biederman J (1997) Neurology 48:589–601.
22. Mostofsky SH, Cooper KL, Kates WR, Denckla MB, Kaufmann WE (2002)
Biol Psychiatry 52:785–794.
23. Hill DE, Yeo RA, Campbell RA, Hart B, Vigil J, Brooks W (2003) Neuropsychology
17:496–506.
24. Casey BJ, Tottenham N, Fossella J (2002) Dev Psychobiol 40:237–254.
25. Tamm L, Menon V, Reiss AL (2002) J Am Acad Child Adolesc Psychiatry
41:1231–1238.
26. Posner MI, DiGirolamo GJ (1998) in The Attentive Brain, ed Parasuraman R
(MIT Press, Cambridge, MA).
27. Rogers RD, Owen AM, Middleton HC, Williams EJ, Pickard JD, Sahakian BJ,
Robbins TW (1999) J Neurosci 19:9029–9038.
28. Ernst M, Bolla K, Mouratidis M, Contoreggi C, Matochik JA, Kurian V, Cadet
JL, Kimes AS, London ED (2002) Neuropsychopharmacology 26:682–691.
29. Fletcher PC, Henson RN (2001) Brain 124:849–881.
30. Castellanos FX, Tannock R (2002) Nat Rev Neurosci 3:617–628.
31. Sowell ER, Thompson PM, Welcome SE, Henkenius AL, Toga AW, Peterson
BS (2003) Lancet 362:1699–1707.
32. Carmona S, Vilarroya O, Bielsa A, Tremols V, Soliva JC, Rovira M, Tomas J,
Raheb C, Gispert JD, Batlle S, et al. (2005) Neurosci Lett 389:88–93.
33. Kaya GC, Pekcanlar A, Bekis R, Ada E, Miral S, Emiroglu N, Durak H (2002)
Ann Nuclear Med 16:527–531.
34. Tamm L, Menon V, Ringel J, Reiss AL (2004) J Am Acad Child Adolesc
Psychiatry 43:1430–1440.
35. Schweitzer JB, Faber TL, Grafton ST, Tune LE, Hoffman JM, Kilts CD (2000)
Am J Psychiatry 157:278–280.
36. Smith AB, Taylor E, Brammer M, Toone B, Rubia K (2006) Am J Psychiatry
163:1044–1051.
37. Mulas F, Capilla A, Fernandez S, Etchepareborda MC, Campo P, Maestu F,
Fernandez A, Castellanos FX, Ortiz T (2006) Biol Psychiatry 59:373–379.
38. Barry RJ, Johnstone SJ, Clarke AR (2003) Clin Neurophysiol 114:184–198.
39. Mesulam MM (1998) Brain 121:1013–1052.
40. Gogtay N, Sporn A, Clasen LS, Nugent TF, III, Greenstein D, Nicolson R,
Giedd JN, Lenane M, Gochman P, Evans A, et al. (2004) Arch Gen Psychiatry
61:17–22.
41. Sowell ER, Thompson PM, Leonard CM, Welcome SE, Kan E, Toga AW
(2004) J Neurosci 24:8223–8231.
42. Chklovskii DB, Mel BW, Svoboda K (2004) Nature 431:782–788.
43. Sur M, Rubenstein JL (2005) Science 310:805–810.
44. Hensch TK (2004) Annu Rev Neurosci 27:549–579.
45. Knudsen EI (2004) J Cognit Neurosci 16:1412–1425.
46. Huttenlocher PR, Dabholkar AS (1997) J Comp Neurol 387:167–178.
47. Russell VA (2007) J Neurosci Methods 161:185–198.
48. Davids E, Zhang K, Tarazi FI, Baldessarini RJ (2003) Brain Res Brain Res Rev
42:1–21.
49. Lenroot RK, Gogtay N, Greenstein DK, Wells EM, Wallace GL, Clasen LS,
Blumenthal JD, Lerch J, Zijdenbos AP, Evans AC, et al. (2007) NeuroImage
36:1065–1073.
50. Luders E, Narr KL, Thompson PM, Woods RP, Rex DE, Jancke L, Steinmetz
H, Toga AW (2005) NeuroImage 26:493–501.
51. Syed Z, Dudbridge F, Kent L (2007) Am J Med Genet 144:B375–B378.
52. Kent L, Green E, Hawi Z, Kirley A, Dudbridge F, Lowe N, Raybould R,
Langley K, Bray N, Fitzgerald M, et al. (2005) Mol Psychiatry 10:939–943.
53. Reich W (2000) J Am Acad Child Adolesc Psychiatry 39:59–66.
54. Werry JS, Sprague RL, Cohen MN (1975) J Abnorm Child Psychol 3:217–229.
55. Giedd JN, Rumsey JM, Castellanos FX, Rajapakse JC, Kaysen D, Vaituzis AC,
Vauss YC, Hamburger SD, Rapoport JL (1996) Brain Res Dev Brain Res
91:274–280.
56. Castellanos FX, Giedd JN, Marsh WL, Hamburger SD, Vaituzis AC, Dickstein
DP, Sarfatti SE, Vauss YC, Snell JW, Lange N, et al. (1996) Arch Gen Psychiatry
53:607–616.
57. Sled JG, Zijdenbos AP, Evans AC (1998) IEEE Transactions on Medical
Imaging 17:87–97.
58. Zijdenbos AP, Forghani R, Evans AC (2002) IEEE Transactions on Medical
Imaging 21:1280–1291.
59. MacDonald D, Kabani N, Avis D, Evans AC (2000) NeuroImage 12:340–356.
60. Lerch JP, Evans AC (2005) NeuroImage 24:163–173.
61. Blumenthal JD, Zijdenbos A, Molloy E, Giedd JN (2002) NeuroImage 16:89–92.
62. Pinheiro JC, Bates DM (2000) Mixed-Effects Models in S, S-PLUS (Springer,
New York).
19654 www.pnas.org cgi doi 10.1073 pnas.0707741104 Shaw et al.