Longitudinal studies have tracked structural brain development across childhood and adolescence using MRI. One study analyzed MRI data from over 33 participants who each underwent multiple scans from ages 7 to 30 years old. The study found that regions involved in social cognition like the temporoparietal junction continued developing structurally into adolescence, with cortical thickness decreasing and surface area peaking then decreasing. Exploring the hypothesis that a mismatch between limbic and prefrontal cortex maturation drives adolescent behaviors, the study aimed to observe developmental patterns longitudinally at the individual level.

1. Measuring changes in brain structure
across childhood and adolescence
Brain & Development Lab, 17 June 2013
Kate Mills

2. Thank You!
Sarah-Jayne Blakemore
Anne-LiseGoddings
Jay Giedd
LivClasen
UCL Developmental Cognitive
Neuroscience Group

3. Outline
• Longitudinal studies of structural brain development
• What does structural MRI measure?
• Developmental changes in the structure of the social brain
in late childhood and adolescence
• Exploring the developmental mismatch hypothesis with
structural brain data
• Assessing brain maturation

4. Structural Brain Imaging
DTI

5. Structural Brain Imaging
Roberto Toro Brain Catalogue

6. Outline
• Longitudinal studies of structural brain development
• What does structural MRI measure?
• Developmental changes in the structure of the social brain
in late childhood and adolescence
• Exploring the developmental mismatch hypothesis with
structural brain data
• Assessing brain maturation

7. Longitudinal studies of structural brain development
Giedd et al., 1999: 145 participants 243 scans
Sowell et al., 2004: 45 participants 90 scans
Gogtay et al., 2004: 13 participants 52 scans
Lenroot et al., 2007: 387 participants 829 scans
Raznahan et al., 2011: 647 participants 1250 scans
Urošević et al., 2012: 149 participants 298 scans
van Soelen et al., 2012: 113 participants 226 scans
Pfefferbaum et al., 2013: 47 participants 112 scans
Tamnes et al., 2013: 85 participants 170 scans
Mutlu et al., 2013: 137 participants 209 scans
Auber-Broche et al., 2013: 292 participants 882 scans
...and more each day!

8. Giedd et al., 1999: 145 participants 243 scans
Sowell et al., 2004: 45 participants 90 scans
Gogtay et al., 2004: 13 participants 52 scans
Lenroot et al., 2007: 387 participants 829 scans
Raznahan et al., 2011: 647 participants 1250 scans
Urošević et al., 2012: 149 participants 298 scans
van Soelen et al., 2012: 113 participants 226 scans
Pfefferbaum et al., 2013: 47 participants 112 scans
Tamnes et al., 2013: 85 participants 170 scans
Mutlu et al., 2013: 137 participants 209 scans
Auber-Broche et al., 2013: 292 participants 882 scans
Child Psychiatry Branch NIMH (Giedd, Rapoport)

9. Longitudinal studies of structural brain development
• Gray matter density
measures the proportion
of gray matter in a small
region of fixed radius (15
mm) around each
cortical point. Similar to
cortical thickness.
• Non-linear changes not
captured in video.
Gray matter maturation over the cortical surface between ages 4 and 21. The process
of gray matter maturation is represented by the blue color. The side bar shows a color
representation in units of gray matter volume.
(Gogtay et al., 2004)

10. Center for the Study of Human Cognition / Lifespan
Changes in Brain and Cognition (Walhovd, Fjell)
Giedd et al., 1999: 145 participants 243 scans
Sowell et al., 2004: 45 participants 90 scans
Gogtay et al., 2004: 13 participants 52 scans
Lenroot et al., 2007: 387 participants 829 scans
Raznahan et al., 2011: 647 participants 1250 scans
Urošević et al., 2012: 149 participants 298 scans
van Soelen et al., 2012: 113 participants 226 scans
Pfefferbaum et al., 2013: 47 participants 112 scans
Tamnes et al., 2013: 85 participants 170 scans
Mutlu et al., 2013: 137 participants 209 scans
Auber-Broche et al., 2013: 292 participants 882 scans

11. (Tamnes et al., 2013)
Longitudinal studies of structural brain development

12. How much does cortical thickness change across
development?
Sowell et al., 2004: 5-11 years 0.3 – 0.6mm
Raznahan et al., 2011: 8-22 years ~.25mm
van Soelen et al., 2012: 9-12 years <.24mm
Mills et al., 2013: 9-22 years 0.1 – 0.4mm
Mutlu et al., 2013: 6-30 years ~.5mm

13. Why longitudinal?
variability between individuals > variability within individuals
Aubert-Broche et al., 2013
Total cerebral volume Total cerebral volume
van Soelen et al., 2013Lenroot et al., 2007
Total cerebral volume

14. Why longitudinal?
variability between individuals > variability within individuals
Aubert-Broche et al., 2013
Gray Matter Volume Cerebral Cortex Volume
Tamnes et al., 2013
Gray Matter Volume
Raznahan et al., 2011

15. Steen et al., 2007

16. Outline
• Longitudinal studies of structural brain development
• What does structural MRI measure?
• Developmental changes in the structure of the social brain
in late childhood and adolescence
• Exploring the developmental mismatch hypothesis with
structural brain data
• Assessing brain maturation

17. 1mm3 cortical voxel
contains:
• 20,000 – 30,000 neurons
• Up to twice as many glial
cells
• 0.4 km dendrites
• 4 km axons
• 400,000,000 to
1,000,000,000 synapses
What does structural MRI measure?
(Cragg, 1967; Logothetis, 2008; Sherwood et al., 2006; Pelvig et al., 2008)

18. What does structural MRI measure?
(Chung et al., 2013)
0.5-mm-thick blocks of BA10 from an autistic patient, stored in
formalin for >6 years. Stained for axons with neurofilament
protein and myelin basic protein to trace individual fibres.

19. (Chung et al., 2013)

20. What does cortical thickness measure?
(Carlo and Stevens, 2013)
Cortical thickness relates to number of glial cells, not number of neurons

21. Does synaptic number affect cortical thickness?
(Bourgeois and Rakic, 1993)

22. Does synaptic number affect cortical thickness?
(Bourgeois and Rakic, 1993)
“Changes in the density of synapses affect very little either the volume or the
surface of the cortex because the total volume of synaptic boutons (the synaptopil
illustrated in Fig. 2) is only a very small fraction of the cortical volume.”
“If we assume that the synaptic contacts are basically spheres (Fig. 2) with this
mean diameter, then they would represent only 2% of 1 mm3 of the neuropil or
less than 1.5% of the cortical volume, even at this exceptionally high density of
synapses. Thus, a decline of synaptic number during puberty should have a rather
small effect on the overall volume of the cortex.”
“Neither the overall percentage of neuropil in the cortex nor the volume of the
cortex itself changes significantly during puberty (present data; R. Williams, K.
Ryder, and P. Rakic, unpublished observations).”
“For example, layers II and III display the highest decrease in synaptic density
although these layers have the most steady percentage of neuropil during the life
span. In contrast, layer VI, in which we have observed the largest decrease in
percentage of neuropil, displays the smallest change in density of synapses.”

23. What does structural MRI measure?
Gray Matter Volume
White Matter Volume
Cortical Thickness
Surface Area
Gyrification/Folding
Cortical Thickness x Surface Area
Ratio of gray relative to white within a sphere
Myelination
Axonal calibre
Synaptic pruning
Intracorticalmyelination
Axonal calibre
Glial loss
Number of cortical columns in a functional area
Mechanistic forces and developmental timing

24. Outline
• Longitudinal studies of structural brain development
• What does structural MRI measure?
• Developmental changes in the structure of the social brain
in late childhood and adolescence
• Exploring the developmental mismatch hypothesis with
structural brain data
• Assessing brain maturation

25. Developmental changes in the structure of the social
brain in late childhood and adolescence
Mentalizingis the ability to
infer the intentions, beliefs
and desires of others in
order to predict their
behavior.
Behavioral and functional
neuroimaging studies
suggest this ability
continues to develop across
the second decade.

26. “Social Brain Network”
mBA10
TPJ
pSTS
ATC

27. Defining the temporoparietal junction
(Mars et al., 2011)

28. Participant Characteristics

29. Image Processing
(Winkler et al., 2010)
Thickness
White surface
Pial Surface
Surface Area
Gray matter
volume
Surface-based Representation
Cortical Thickness Surface Area = Gray Matter Volume

30. Statistical Analysis
Mixed-modeling and AIC to determine the best fitting
model (cubic, quadratic, or linear)
Gray matter
measurement
Age

32. • Regions of the social brain continue to develop structurally
across adolescence.
• Cortical thickness decreases in the TPJ, pSTS, and mBA10
across adolescence, whereas the ATC increases in cortical
thickness until late adolescence.
• Surface area for each region followed a cubic trajectory,
peaking in early or pre-adolescence before decreasing into the
early twenties.
• Sex differences in gray matter volume appear to be driven by
surface area.
• Age differences in gray matter volume are a product of changes
in surface area and cortical thickness.
Conclusion

33. red = constant
green = linear
blue = quadratic
orange = cubic
Cortical thickness development types

34. Sex differences in cortical thickness development
Thinning

35. Outline
• Longitudinal studies of structural brain development
• What does structural MRI measure?
• Developmental changes in the structure of the social brain
in late childhood and adolescence
• Exploring the developmental mismatch hypothesis with
structural brain data
• Assessing brain maturation

36. Background
• Regions of the human brain develop at different rates across the
first two decades of life.
• Multiple functional imaging studies show heightened recruitment
of limbic structures in adolescents compared to adults.
• It has been hypothesized that a mismatch in the timing of
maturation between limbic structures (i.e., nucleus accumbens
and amygdala) and the prefrontal cortex may underlie some
adolescent behaviors.
From Somerville, Jones & Casey, 2010

37. • Regions of the human brain develop at different rates across the
first two decades of life.
• Multiple functional imaging studies show heightened recruitment
of limbic structures in adolescents compared to adults.
• It has been hypothesized that a mismatch in the timing of
maturation between limbic structures (i.e., nucleus accumbens
and amygdala) and the prefrontal cortex may underlie some
adolescent behaviors.
• Most support for this hypothesis relies on cross-sectional data.
• It is not known if this pattern can be observed on an individual
level.
Background

38. Participant Characteristics
No. of participants: 33
No. of scans: 152
Age range: 7.01-29.9
Gender: 10 Female, 23 Male
IQ: 118 11
scans
scans
Scans per participant
3
4
5
6
scans
scans

39. Methods
Subcortical
Segmentation
Surface-based
Cortical
Reconstruction
• All participants had at least three high quality scans across
adolescence.
• FreeSurfer5.3 (!!) longitudinal pipeline.
• Measures of gray matter volume were obtained for amygdala,
nucleus accumbens and prefrontal cortex.
• Non-linear mixed-modeling was implemented using the nlme
package in R to determine the best fitted model (cubic, quadratic,
or linear).

40. Prefrontal Cortex
Nucleus Accumbens
Amygdala
Regions of Interest

41. cubic p<.0001 linear p<.0001 quadratic p<.0001
Best fitting models across all participants
Raw volumes for all participants

42. Models scaled over raw volumes

43. Findings from other studies
(Tamnes et al., 2013)

44. Findings from other studies
Early Adolescent
9-12 years
Late Adolescent
13-17 years
Young Adult
18-23 years
( et al., 2012)

45. Findings from other studies
(Dennison et al., 2013)
between ages ~12.5 and ~16.5

46. Findings from other studies
(Mohr and Sisk, 2013)
Pubertally-born neurons and glial cells in the medial
amygdala.

47. (Gee et al., 2013)
Findings from other studies

48. (Gee et al., 2013)
Findings from other studies

49. (Christakou et al., 2013)
Findings from other studies

50. (Costa Dias et al., 2013)
Findings from other studies

51. Prefrontal cortex subdivisions

52. Prefrontal cortex subdivisions
cubic p<.0001 cubic p<.02 cubic p<.02

53. Structural stability as maturational index
(Pfefferbaum et al., 2013)

54. age
160000
180000
200000
10 16 22 28
Volume
0.90
1.00
1.10
10 16 22 28
Proportionoffinal
timepointvolume
0.90
1.00
1.10
10 16 22 28
0.90
1.00
1.10
10 16 22 28
0.90
1.00
1.10
10 16 22 28
Maturational graphs for each participant

55. Maturational graphs for each participant

56. • The developmental mismatch hypothesis between the
nucleus accumbens, amygdala and prefrontal cortex
appears to be supported by longitudinal structural data.
• The rate of structural volume change starts to decrease in
mid- to late adolescence for the amygdala, whereas the
prefrontal cortex and nucleus accumbens show continual
change into the mid-twenties.
• This temporal mismatch in structural development is
observable to a variable extent between individuals.
Conclusion

57. Outline
• Longitudinal studies of structural brain development
• What does structural MRI measure?
• Developmental changes in the structure of the social brain
in late childhood and adolescence
• Exploring the developmental mismatch hypothesis with
structural brain data
• Assessing brain maturation

58. “Converging neuropsychology, brain imaging and electrical activity evidence
suggests that breastfed infants may display preferential myelinationand white
matter development.”
“improved developmental growth in late maturing white matter association
regions”
“While prior imaging studies have shown increased brain volume and cortical
thickness in adolescents who were breastfed as infants (Hallowell and
Spatz, 2012; Isaacs et al., 2010; Kafouri et al., 2012)”

59. (Deoni et al., in press)

60. The age of attaining peak cortical thickness in children with ADHD compared with
typically developing children

61. Structural development, an issue of timing?
• When is “less/more” better?
• What do trajectory rates mean?
• Does cortical thickness mean the same thing at different
stages of development?
• How does all of this relate to cognitive/behavioral
development?

62. “We interpret these data
to suggest that the NAcc
development may
precede that of the OFC
during adolescence.
Protracted development
of prefrontal regions, with
a transition from diffuse
to focal recruitment is
consistent with MRI-
based neuroanatomical
[studies]”

64. (Dosenbach et al., 2010)

65. BRAINSTORM