11008_The Worsening Trajectory Of Social Impairment In Preterm Born Young Adults And Its Association

Yale University
EliScholar – A Digital Platform for Scholarly Publishing at Yale
Yale Medicine Thesis Digital Library
School of Medicine
January 2019
The Worsening Trajectory Of Social Impairment In
Preterm Born Young Adults And Its Association
With Altered Amygdalar Functional Connectivity
Christina Johns
Follow this and additional works at: https://elischolar.library.yale.edu/ymtdl
This Open Access Thesis is brought to you for free and open access by the School of Medicine at EliScholar – A Digital Platform for Scholarly
Publishing at Yale. It has been accepted for inclusion in Yale Medicine Thesis Digital Library by an authorized administrator of EliScholar – A Digital
Platform for Scholarly Publishing at Yale. For more information, please contact elischolar@yale.edu.
Recommended Citation
Johns, Christina, “The Worsening Trajectory Of Social Impairment In Preterm Born Young Adults And Its Association With Altered
Amygdalar Functional Connectivity” (2019). Yale Medicine Thesis Digital Library. 3506.
https://elischolar.library.yale.edu/ymtdl/3506

The worsening trajectory of social impairment in preterm born young adults and its
association with altered amygdalar functional connectivity

A Thesis Submitted to the Yale University School of Medicine in Partial Fulfillment
of the Requirements for the Degree of Doctor of Medicine

by
Christina B. Johns
2019

THE WORSENING TRAJECTORY OF SOCIAL IMPAIRMENT IN PRETERM
BORN YOUNG ADULTS AND ITS ASSOCIATION WITH ALTERED
AMYGDALAR FUNCTIONAL CONNECTIVITY

Christina B. Johns1, Cheryl Lacadie2, Betty Vohr3, Dustin Scheinost2, Laura R. Ment1,4.
1Department of Pediatrics, Yale University School of Medicine, New Haven, CT, USA, 2Department of Radiology and
Biomedical Imaging, Yale University School of Medicine, New Haven, CT, USA, 3Department of Pediatrics, Warren
Alpert Medical School of Brown University, Providence, RI, USA and 4Department of Neurology, Yale University
School of Medicine, New Haven, CT, USA.

Survivors of preterm birth experience long-lasting behavioral problems characterized
by increased risk of depression, anxiety, and impaired social functioning. The amygdala
is a key region for social functioning, and alterations in amygdalar structure and
connectivity are thought to underlie social functioning deficits in many disorders,
including preterm birth. However, the trajectory of social impairments in PT and their
association with functional connectivity of the amygdala are not well-studied in former
preterm born individuals (PTs).
It was hypothesized that PTs would show impaired social functioning compared to
term controls beginning in early childhood and continuing to young adulthood. It was
also hypothesized that amygdala resting state functional connectivity is altered in PT born
young adults, and that alterations in amygdala functional connectivity would mediate
increased internalizing behavior and socialization problems in PT born young adults.
In a group of former very PT infants (600 to 1250 grams birth weight) and matched
term (T) controls, measures of social and emotional behavior were examined using the
Child Behavior Checklist (CBCL) administered at ages 8, 12, and 16, the Youth Self
Report administered at age 16, and the Vineland Adaptive Behavior Scales (VABS)
administered at ages 8 and 18. Amygdalar functional connectivity was examined using
resting-state functional magnetic resonance imaging at age 20.
By parent report, preterm-born children and adolescents exhibit behaviors
demonstrating increased social impairment compared to their term-born peers, starting at
school-age and becoming more prominent by young adulthood. PT demonstrate a
worsening trajectory in CBCL Withdrawn scores from school-age to young adulthood
compared to T (group*time interaction p=0.03), and maternal education has a protective
effect on this trajectory in the PT population (withdrawn group*time interaction p=0.01).
Furthermore, amygdalar connectivity is altered in the formerly prematurely-born, and
markers of social impairment correlate negatively with altered amygdala-posterior
cingulate cortex connectivity (Social competence r=-0.37, p=0.03; socialization r=-0.42,
p=0.01).
As this cohort of PTs does not include individuals who suffered any form of
neurologic injury, their parent-reported increase in behavioral markers of social
impairment may be attributable to prematurity rather than to neurologic injury. Moreover,
these data suggest that previously established social impairments in PT as compared to T
worsen during the critical period of transition from school-age to adolescence and suggest
a possible neural underpinning for these impairments experienced by prematurely-born
individuals.

Acknowledgements

I thank Dr. Laura Ment for her guidance and encouragement over the last four years. She
has taught me much about preterm neurodevelopment, research design, and balancing a
research and clinical career and I’m very grateful for her mentorship.

I also thank Dr. Dustin Scheinost for his ideas and guidance which were central to the
completion of this work and for his assistance in writing up the original manuscript.

Thank you to Dr. Betty Vohr for her insights during the completion of this analysis and to
Cheryl Lacadie for her assistance with the fMRI analyses.

I thank the following individuals for their participation in the original collection of data
used in this work: Drs. Deborah Hirtz and Walter Allan for their scientific expertise;
Marjorene Ainley for the follow-up coordination; Jill Maller-Kesselman, Susan Delancy
and Victoria Watson for their neurodevelopmental testing; Hedy Sarofin and Terry
Hickey for their technical assistance.

Finally, I thank the children and their families for their participation in the study.

This work was supported by NIH NS27116 and by the Vernon W. Lippard MD Student
Summer Research Fellowship.

Table of Contents
Table of Frequently Used Abbreviations
…………………………………………………………………. 1
Introduction
…………………………………………………………………………………………………………. 2
Specific Hypotheses and Aims ………………………………………………………………………………. 7
Methods………………………………………………………………………………………………………………. 8
Results
………………………………………………………………………………………………………………. 17
Discussion …………………………………………………………………………………………………………. 39
References
…………………………………………………………………………………………………………. 48

1

Table of Frequently Used Abbreviations
PT
Preterm
T
Term
CBCL
Child Behavior Checklist
YSR
Youth Self Report
VABS
Vineland Adaptive Behavior Scales
rs-fMRI
Resting state functional magnetic resonance imaging
PCC
Posterior cingulate cortex
L-STG
Left superior temporal gyrus

2
Introduction1
Premature Birth: Overall Implications
Emerging data suggest that preterm-born children are at high risk for social
impairment and emotional problems in addition to the well-established risk of
neurodevelopmental handicap; however, the latter is much more well-described and
remains largely the focus of counseling families about the longterm risks to prematurely
born individuals.
Preterm birth is a significant global public health problem: in 2017, 9.93% of US
births were preterm, with 2.76% born before 34 weeks (2). Globally, as many as 11% live
births occur before 37 weeks of gestation (3, 4). In the US, the rate of PT birth increased
from the 1980s through 2006 and has recently begun increasing again over the last few
years (5). There are racial, ethnic, and socioeconomic disparities in rates of preterm birth,
with non-Hispanic African Americans having the highest rates and even higher rates
among mothers with low educational attainment (6).
The consequences of preterm birth are far-reaching and include acutely increased
mortality as well as significant long-term morbidity and increased societal costs.
Advances in obstetric and neonatal care have improved survival for preterm born
neonates; however, these children are still at high risk for significant health problems,
including physical as well as neurodevelopmental problems (6, 7). These include
pulmonary and cardiovascular problems, major neurologic impairments such as cerebral
palsy, cognitive impairment, and sensory impairments, and more subtle learning,

1 Portions of thesis text are taken from the author’s published manuscript:
1.
Johns CB, Lacadie C, Vohr B, Ment LR, and Scheinost D. Amygdala functional
connectivity is associated with social impairments in preterm born young adults.
Neuroimage Clin. 2018.

3
behavioral, and emotional problems (6, 8-10). In 2010, about 2.7% of PT survivors
globally were estimated to have moderate or severe neurodevelopmental impairments,
and the number of PT survivors with subtler emotional or behavioral problems is likely
much higher though not well established (3).
Emotional and Social Problems in the Prematurely Born
Survivors of preterm birth experience long-lasting behavioral problems
characterized by increased risk for depression, anxiety, and impairments in social
functioning (8-13). Social difficulties in PT emerge in early childhood and persist into
adolescence. In early childhood, PTs show increased internalizing behavior, impaired
emotional regulation, and poorer peer play, and are reported by parents to have increased
social problems (14-17). Specific domains in which PT commonly struggle compared to
T include social withdrawal and difficulties with peers (18).
The transition to adolescence appears to be especially difficult for PTs. A recent
prospective study of behavioral and emotional problems in extremely PT-born children
from school-age to young adulthood showed consistent increase in emotional symptoms
and peer problems in PT compared to T controls which was greater in young adulthood
compared to school-age (19). This is concordant with an increased risk of bullying in PT
in adolescence (20, 21). Furthermore, PTs show increased internalizing behaviors both by
parent and teacher report in early adolescence (22) and fail to follow the age-related
normal decline in these behaviors during the transition from adolescence to adulthood
(23). It is theorized that decreased social skills in early childhood and a rise in
internalizing behaviors may lead to difficult social relationships in adolescence and
young adulthood in PT, which then manifests as social withdrawal (18).

4
Even in adulthood, PT are less extraverted, take fewer risks, and have lower self-
esteem compared to their term-born peers (12, 24). Because of these impairments in
social functioning, PT-born adults are less likely to maintain committed relationships or
become parents (25). In addition, these symptoms have been linked to increased
psychiatric morbidity in the PT population at young adulthood, including anxiety,
depression, and social phobias (10, 11, 26-28). Interestingly, most of these reports are
from parents or caregivers, and self-report data are rarer. However, in general, even when
parents report social, emotional, and behavioral problems, PT-born adolescents do not
report significant problems compared to term peers (29, 30).
Neurodevelopment in Prematurely Born Individuals
Preterm birth is associated with alterations in cortical and subcortical regional
volume as well as with disruptions in neural connectivity networks that can persist into
adolescence and adulthood (31-33). While some of these changes may be due to perinatal
factors including procedures (34) during what would normally be a period of significant
neurodevelopment while in utero (35), there is increasing evidence that pre-natal factors
such as maternal stress may play a role (36, 37). While many cortical and subcortical
areas may be affected by preterm birth, the limbic areas are of particular interest given
their role in responding to stress and coordinating emotional responses.
The Amygdala: Function and Connectivity

A key brain region for social functioning is the amygdala (38). Lesion studies
show that damage to the amygdala impairs individuals’ abilities to recognize complex
social emotions in facial expressions (39, 40). Amygdalar volume and functional
connectivity with cortical regions correlates with social network size in young adults (41,
42), and alterations to amygdalar circuitry contribute to social processing deficits in many

5
disorders, such as autism spectrum and anxiety disorders (43-45). Similarly, reduced
social functioning in PTs has been attributed to alterations in amygdalar structure and
function (13, 46-48).
The amygdala develops early in life and exhibits some volume and connectivity
changes from infancy to adulthood in typically developing individuals. The amygdala
grows rapidly during infancy in healthy full-term born children and reaches its maximum
volume by late school-age, with small volume changes during adolescence and adulthood
(49, 50). Amygdalar functional connectivity develops similarly early in life: in healthy
full-term infants, the amygdala is positively correlated with subcortical regions including
the contralateral amygdala, hippocampus, insula, hypothalamus, and thalamus and
negatively correlated with the prefrontal cortex, posterior cingulate cortex, and visual
cortex (36, 46). In late infancy and early childhood, amygdalar-thalamic connectivity
decreases and amygdalar-right ventral temporal lobe connectivity increases (51), but
from early childhood to adulthood, amygdalar connectivity with subcortical regions
remains largely unchanged with the exception of a few regions (52). Amygdalar
connectivity with the medial prefrontal cortex increases with age beginning around age
10, whereas connectivity with a region including the insula and superior temporal sulcus
as well as with the posterior cingulate cortex decreases with age after early adolescence
(52). Additional subtle amygdalar connectivity changes are mediated by both post-natal
factors such as parental interactions (53-55) and pre-natal factors including maternal
stress (36, 37) with potential subsequent consequences for emotional and social
development.
While alterations in functional connectivity for specific networks, such as
language, are well characterized across development in those prematurely born (32),

6
functional connectivity of the amygdala in PT has been less well-studied. In PT neonates,
amygdalar connectivity is decreased to frontal cortex and sub-cortical regions (36, 46)
and correlates with internalizing symptoms at 2 years of age (46). In PT adults at 30
years of age, amygdalar connectivity is decreased to the right posterior cingulate cortex,
left precuneus, and increased to the superior temporal sulcus (47). However, despite
evidence that amygdalar connectivity in typically developing individuals exhibits
changes during adolescence and young adulthood (52, 56), this age range has not been
examined in previous studies of amygdalar connectivity in PTs. Together, these studies
suggest the need to investigate the association between social functioning and amygdalar
connectivity in PT young adults.
In this work, we examined social functioning from school age to young adulthood
and amygdalar connectivity during young adulthood in a cohort of very PT and term
control participants. Measures of social and emotional development were evaluated by
both parent and self-report at ages 8, 12, 16 and 18. Neuropsychological scores were
examined longitudinally for both PT and T. Assessment scores were then compared to
amygdalar functional connectivity using resting-state functional magnetic resonance
imaging between study groups at age 20, and finally, social behavior differences were
correlated with alterations in the amygdala.

7
Specific Hypotheses and Aims
Hypotheses:
Hypothesis 1: Preterms without any history of perinatal brain injury will show
significantly more internalizing behavior and social difficulties beginning at age 8
compared to term-born peers, and these difficulties will persist into young adulthood.
Hypothesis 2: Resting fMRI patterns of amygdala – cortical connectivity will differ
between term and preterm born young adults.
Hypothesis 3: Alterations in functional connectivity will correlate with increased
internalizing behavior and socialization problems seen in adolescents and young adults
who were born preterm.
Specific Aims
Specific Aim 1: To further clarify the trajectory of internalizing behavior and social
problems from school-age to young adulthood in preterms without any significant history
of perinatal brain injury.
Specific Aim 2: To elucidate the development of amygdala – cortical
functional connectivity in adolescents and young adults who were born preterm.
Specific Aim 3: To correlate those connectivity differences with differences in
internalizing behaviors and socialization problems in children and adolescents born
preterm vs. full-term.

8
Methods

This study was designed by Christina Johns, Laura Ment, MD, and Dustin
Scheinost, PhD. The neuropsychological data and rs-fMRI data were collected as part of
the follow-up MRI component of the Multicenter Randomized Indomethacin
Intraventricular Hemorrhage Prevention Trial (NS27116), which was designed and led by
Dr. Ment and performed at the Yale University School of Medicine in New Haven, CT,
the Warren Alpert Medical School of Brown University in Providence, RI, and Maine
Medical Center in Portland, ME (57, 58). The protocols for this study were reviewed and
approved by institutional review boards at each study center. Children provided written
assent; parent(s) or guardians provided written consent for the study. Brain scans were
obtained and analyzed at the Yale University School of Medicine.

Statistical analyses of the neuropsychological data were designed by Christina
Johns with the guidance of Drs. Ment and Scheinost. Analyses of the rs-fMRI data were
designed by Dr. Scheinost as described in detail by him below (see Image Parameters,
Common Space Registration, Connectivity Processing, Amygdalar Seed Connectivity,
and Motion Analysis below). Rs-fMRI analysis was performed by Christina Johns, Dr.
Scheinost, and Cheryl Lacadie. Connectivity and neuropsychological correlations were
performed by Christina Johns.
Participants
The PT neuropsychological cohort consisted of the 437 surviving former PT
participants enrolled in the follow-up MRI component of the Multicenter Randomized
Indomethacin Intraventricular Hemorrhage Prevention Trial (57, 58). The PT
participants all weighed between 600-1250 grams at birth. These participants were
evaluated at ages 8, 12, 16 and 18 with neuropsychological testing. At each age point,

9
PTs were excluded from the neuropsychological analysis for any of three reasons: 1. Any
evidence of perinatal brain injury, defined by intraventricular hemorrhage, low-pressure
ventriculomegaly, and/or periventricular leukomalacia, 2. Incomplete demographic,
WISC, or neuropsychological questionnaires, and 3. Outlier scores on any of the included
neuropsychological measures. Outlier scores were defined as scores at least 3
interquartile range above the third quartile on any of the included measures.
A subset of participants recruited from the Yale site only was tested with the
Youth Self Report (YSR) at age 16. Participants were excluded from analysis of this
questionnaire for the same reasons as above.
Term (T) control participants were recruited at age 8 years from the local
community or randomly selected from a telemarketing list and matched to the PT
participants in terms of age, gender, and zip code, as a proxy for socio-economic status.
Term controls participated in the 8, 12, 16, and 18-year visits.
A subset of participants from the neuropsychological cohort was recruited for
MRI testing at age 20 years.
Neuropsychological Assessment

All participants were tested with the CBCL (59) at ages 8, 12, and 16 years and
the VABS (60) at ages 8 and 18 years to assess social and emotional development and
adaptive behavior. Participants also completed the Weschler Intelligence Scale for
Children, Third Edition (WISC-III) (61) at ages 8, 12, and 16 years to assess intellectual
ability, from which Full IQ (FIQ) scores were used in the analysis. A subset of
participants was tested with the YSR (62) at age 16 years to assess social and emotional
development from the participant’s, rather than the parent’s, point of view. T scores for
each domain were used for the CBCL, YSR, and VABS.

10
The CBCL is a validated, parent/caregiver-completed questionnaire of child
emotional and behavioral problems over the past 6 months. Measures of social
development included in this study included scores in the following scales: Social
Competence, Social Problems, Anxiety Problems, Anxious/Depressed, Withdrawn, and
Affect Problems. At ages 8 and 12 years, only the Social Problems, Anxious/Depressed,
and Withdrawn scales were assessed. In this questionnaire, higher scores for Social
Problems, Anxiety Problems, Anxious/Depressed, Withdrawn, and Affect Problems
reflect a worse level of functioning, whereas lower scores in Social Competence reflect a
worse level of functioning. The Social Competence scale includes items such as
participation in activities and frequency of contact with friends, and the Social Problems
scale includes items such as a child’s ability to get along with peers, amount of play time
spent with peers of same age, and whether a child acts his/her age. The Withdrawn scale
includes items such as avoiding eye contact and refusing activity, the Anxious/Depressed
scale includes items such as frequency that the child’s feelings are hurt, whether the child
is upset by separation, and frequency of sadness. The Anxiety Problems scale assesses
dependency, not sleeping alone, and number of fears. Clinical range scores for these
scales are defined as being in the bottom two percentiles of T scores for Social
Competence (T scores £ 37) and the top two percentiles for the remainder of the scales (T
scores ³ 70).
The YSR is similar to the CBCL, but is self-administered (62). Measures from
this instrument included in this study include the following: Activities and Social
(subscales) and Anxious/Depressed, Withdrawn, and Social Problems (syndrome scales).
DSM Affective Problems and DSM Anxiety Problems scales were also included. These
scales assess items similar to those assessed in the CBCL. These DSM-oriented scales are

11
comprised of measures consistent with DSM-5 categories (Affective Problems:
dysthymia and major depressive disorder; Anxiety Problems: generalized anxiety
disorder, separation anxiety, and specific phobia) as identified by experts (63). Clinical
range scores on the YSR are defined as in the CBCL: for the Syndrome and DSM-
oriented scales, scores ³ 70 are in the clinical range, and for the subscales scores £ 31 are
in the clinical range.
The VABS is a parent/caregiver-completed questionnaire that evaluates adaptive
and maladaptive behavior in children. Measures of social development used from the
VABS included scores in the following domains: Adaptive Behavior, Socialization,
Interpersonal Relationships, Play and Leisure Time, and Coping Skills. The latter three
scales are subsets of the “socialization” scale in the VABS. Items assessed in each
domain include the following: Socialization – amount of time playing with peers, helping
others, and sharing toys/possessions, Interpersonal Relationships – asking others to play
and taking turns in activities, Play and Leisure Time – playing in games and playing with
peers, and Coping Skills – controlling anger during unexpected events and cooperation
with others. The Adaptive Behavior domain is a composite measure of the above
domains. At age 8 years, only the Adaptive Behavior and Socialization domains were
assessed. A higher score reflects a better level of function in that domain. Scores £70 for
the Adaptive Behavior and Socialization domains and £10 for the Interpersonal, Play and
Leisure, and Coping domains are designated as clinical range.
Image parameters
Participants were scanned in a Siemens 3T Tim Trio scanner as previously
described at age 20. After a first localizing scan, a high-resolution 3D volume was
collected using a magnetization prepared rapid gradient echo (MPRAGE) sequence (176

12
contiguous sagittal slices, slice thickness 1mm, matrix size 192×192, FoV = 256mm, TR
= 2530 ms, TE = 2.77 ms, flip angle = 7°). Next, a T1-weighted anatomical scan (TR =
300 ms, TE = 2.55 ms, FoV = 220 mm, matrix size 256×256, thickness = 6 mm thick,
gap = 1mm) was collected with 25 AC-PC aligned axial-oblique slices. After these
structural images, acquisition of functional data began in the same slice locations as the
axial-oblique T1-weighted 2D Flash image. Functional images were acquired using a
T2* sensitive gradient-recalled single shot echo-planar pulse sequence (TR = 1550ms, TE
= 30ms, flip angle = 80, Bandwidth = 2056 Hz/pixel, 64*64 matrix, field of view:
220mm x 220mm, interleaved acquisition). Two functional runs consisted of 190
volumes (5-minute scan length) with the first four volumes discarded to allow the
magnetization to reach the steady-state.
Common Space Registration

First, anatomical images were skull stripped using FSL
(https://fsl.fmrib.ox.ac.uk/fsl/) and any remaining non-brain tissue was manually
removed. All further analyses were performed using BioImage Suite (64) unless
otherwise specified. Anatomical images were linearly aligned to the MNI brain using a
12-parameter affine registration by maximizing the normalized mutual information
between images. Next, anatomical images were non-linearly registered to an evolving
group average template in an iterative fashion using a previously validated algorithm.
This algorithm iterates between estimating a local transformation to align individual
brains to a group average template and creating a new group average template based on
the previous transformations. The local transformation was modeled using a free-form
deformation parameterized by cubic B-splines. This transformation deforms an object by
manipulating an underlying mesh of control points. The deformation for voxels in

13
between control points was interpolated using B-splines to form a continuous
deformation field. Positions of control points were optimized using a conjugate gradient
descent to maximize the normalized mutual information between the template and
individual brains. After each iteration, the quality of the local transformation was
improved by increasing the number of control points and decreasing the spacing between
control points to capture a more precise alignment. A total of 5 iterations were performed
with decreasing control point spacings of 15 mm, 10 mm, 5 mm, 2.5, and 1.25 mm. To
help prevent local minimums during optimization, a multi-resolution approach was used
with three resolution levels at each iteration. The functional data were linearly registered
to the 2D Flash image. The 2D Flash image was linearly registered to the MPRAGE
image. All transformation pairs were calculated independently and combined into a single
transform, warping the single participant results into common space. This single
transformation allows the individual participant images to be transformed to the common
space with only one transformation, thereby reducing interpolation error.
Connectivity Processing
Images were slice time and motion corrected using SPM8
(http://www.fil.ion.ucl.ac.uk/spm/). Several covariates of no interest were regressed from
the data, including linear and quadratic drifts, mean cerebral-spinal-fluid (CSF) signal,
mean white-matter signal, and mean gray matter signal. For additional control of possible
motion-related confounds, a 24-parameter motion model (including six rigid-body motion
parameters, six temporal derivatives, and these terms squared) was regressed from the
data. The functional data were temporally smoothed with a Gaussian filter (approximate
cutoff frequency=0.12Hz). A gray matter mask was applied to the data, so only voxels in
the gray matter were used in further calculations.

14
Amygdalar Seed Connectivity
A seed comprised of the bilateral amygdala was defined for the connectivity
analyses (shown in Figure 5) on the reference brain and transformed back (via the inverse
of the transforms described above) into individual participant space. To account for
possible drop-out effect and poor amygdala coverage in the fMRI scans, the overlap
between the amygdala seed and individual participant space was calculated, and
participants with less than 30% overlap were excluded (6 PT and 4 T were excluded from
the analysis based on this). The time course of the reference region in a given participant
was then computed as the average time course across all voxels in the reference region.
This time course was correlated with the time course for every other voxel in gray matter
to create a map of r-values, reflecting seed-to-whole-brain connectivity. These r-values
were transformed to z-values using Fisher’s transform, yielding a map representing the
strength of correlation with the seed for each participant. Finally, the connectivity maps
were smoothed with a 6 mm full width half maximum Gaussian kernel.
Motion Analysis
As group differences in motion have been shown to confound connectivity
studies, we calculated the average frame-to-frame displacement for each participant’s
data. In line with current reports, one PT with an average frame-to-frame displacement
>0.30 were removed from the analysis. We detected no significant difference between
PTs and Ts (PTs: motion=-0.14±0.07; Ts: motion=0.11±0.04; p>0.05).
Statistical Analyses
We analyzed differences in demographic characteristics between PT and T using
Fisher’s exact test for categorical variables and t test for continuous variables.
Demographic variables included gender (reported by the participant at each age point and

15
classified as male or female), maternal education, and race/ethnicity. Maternal education
was classified in a binary fashion as less than a high school education or greater than or
equal to a high school education, and race/ethnicity was classified as White or non-
White.
Linear regression was used to compare neuropsychological outcomes between
PTs and Ts at each age, with covariate adjustment for age at instrument administration,
gender, race/ethnicity, maternal education status, instrument respondent, and full IQ.
Significance was assessed at p<0.05. Repeated measures ANOVA was used to analyze neuropsychological outcomes longitudinally. For these analyses, only subjects with complete testing at ages 8, 12, and 16 (for CBCL measures) and ages 8 and 18 (for VABS measures) were included. Repeated measures ANOVA was also used in a secondary, exploratory analysis to assess the effect of maternal education level on CBCL and VABS scores over time in PT individuals. For the purposes of this analysis, maternal education was classified in a binary fashion as less than a high school education or greater than or equal to a high school education. There were not enough T subjects with complete data to further stratify by maternal education. Pearson’s correlation coefficients were used to assess associations of CBCL and VABS measures over time for PT participants with complete neuropsychological data at each age. There were not enough T subjects with complete neuropsychological data at all ages to perform a correlation analysis. Significance was assessed for each of these analyses at p<0.05. Imaging data were analyzed using voxels t-tests. Significance was assessed at a cluster-level threshold of p<0.01 family-wise error correction for between group comparisons. All maps were corrected for multiple comparisons across gray matter using 16 cluster-level correction estimated via Monte Carlo simulations. AFNI's 3dClustSim (version 16.3.05 which fixed the 3dClustSim “bug”) was used to estimate a cluster size of 1701 mm3 using 10,000 iterations, an initial p-value threshold of 0.01, the gray matter mask using in preprocessing, and smoothness values estimated from the residuals using 3dFWHMx. Exploratory analyses were performed in the sub-cohort of imaged participants to assess the association between functional connectivity and behavior using Pearson’s correlation coefficients. This analysis was restricted to only brain regions and social behavior scores that differed significantly between PTs and Ts in the full behavioral cohort. Additionally, associations were tested within the PT and T groups separately in order to minimize bias. The significance level was p<0.05. 17 Results Participants The PT neuropsychological cohort consisted of the 437 surviving former PT participants enrolled in the follow-up MRI component of the Multicenter Randomized Indomethacin Intraventricular Hemorrhage Prevention Trial (57, 58). Figure 1 details the number of participants included in the analysis at each age point. Figure 1A. Participants included in neurobehavioral analyses at each age. All participants were drawn from the 437 surviving former PT participants enrolled in the follow-up MRI component of the Multicenter Randomized Indomethacin Intraventricular Hemorrhage Prevention Trial. Questionnaire data required for inclusion were a demographic questionnaire and the WISC-III at all age points and the CBCL and VABS at age 8, CBCL at age 12, CBCL at age 16, and VABS at age 18. Outliers were defined as participants scoring at least 3 times the interquartile range above the third or below the first quartile for any of the neurobehavioral outcome measures assessed. 18 At age 8, 199 PTs were included in the analysis of Child Behavior Checklist (CBCL) and Vineland Adaptive Behavior Scales (VABS) testing. 238 participants were excluded from analysis: 62 were lost to follow-up, 100 had evidence of perinatal brain injury, and 61 were excluded due to incomplete testing on the Weschler Intelligence Scale for Children (WISC), CBCL, VABS, or demographic questionnaires. An additional 15 PTs with outlier scores on included measures in the CBCL and/or VABS were excluded from the analysis. The participants who were lost to follow-up at age 8 and who had available demographic data were similar to the included participants in gender makeup and race, but had significantly lower maternal education levels (percentage of participants with maternal education < high school: 11% in included group, 32% in lost to follow-up group, p=0.0002). At age 12, 211 PTs were included in the CBCL analysis. 226 were excluded: 62 were lost to follow-up, 102 had evidence of perinatal brain injury, and 52 had incomplete testing on the WISC, CBCL, or demographic questionnaires. An additional 10 PTs with Figure 1B. Participants included in the YSR analysis at age 16 years. These participants were drawn from the 437 surviving former PT participants enrolled in the follow-up MRI component of the Multicenter Randomized Indomethacin Intraventricular Hemorrhage Prevention Trial, but were only recruited from the Yale site. Questionnaire data required for inclusion were a demographic questionnaire, the WISC-III, and the YSR. Outliers were defined as participants scoring at least 3 times the interquartile range above the third or below the first quartile for any of the neurobehavioral outcome measures assessed. 19 outlier CBCL scores were excluded. Again, the participants who were lost to follow-up were similar in gender and race to the included participants but had significantly lower maternal education levels (percentage of participants with maternal education < high school: 9% in included group, 39% in lost to follow-up group, p<0.0001). At age 16, 161 PTs were included in the analysis. 276 participants were excluded: 100 were lost to follow-up, 86 had evidence of perinatal brain injury, and 89 had incomplete testing on the WISC, CBCL or demographic questionnaires. One PT was labeled as an outlier based on CBCL scores and excluded from analysis. Participants who were lost to follow-up at the 16-year visit were similar in gender and race but had lower maternal education levels (percentage of participants with maternal education < high school: 8% in included group, 36% in lost to follow-up group, p<0.0001). 45 PTs (all recruited from the Yale site only) were included in the YSR analysis at age 16. From the full cohort of PT, 100 participants were excluded due to being lost to follow up, 86 had perinatal brain injury, and 193 were not tested with the YSR. An additional 3 subjects were excluded due to having incomplete WISC or demographic questionnaires. 10 subjects with outlier YSR scores were excluded from the analysis. At age 18, 191 PTs were included in the analysis. Of the 245 participants who were excluded from analysis, 143 were lost to follow-up, 75 had evidence of perinatal brain injury, and 28 had incomplete testing on the WISC, VABS or demographic questionnaires. There were no PTs excluded due to outlier scores on the VABS. The PTs who were lost to follow-up at the 18-year visit were similar in gender makeup to the included PTs but had significantly higher proportions of minority participants (25% included, 42% lost to follow-up, p=0.003) and of participants with low maternal education levels (10% included, 33% lost to follow-up, p<0.0001). 20 At age 8, 25 Ts were included in the CBCL and the VABS analysis, after excluding 18 participants for incomplete questionnaires and 9 for outlier scores. At age 12, 90 Ts were included in the CBCL analysis after excluding 17 participants for incomplete questionnaires and 4 for outlier scores. At age 16, 66 Ts were included in the CBCL analysis, after excluding 27 participants for incomplete questionnaires and 9 for outlier scores. Also at age 16, 56 Ts were included in the YSR analysis, after excluding 41 participants for incomplete data and 5 for outlier scores. At age 18, 71 Ts were included in the VABS analysis, after excluding 10 participants for missing data and 14 participants for outlier scores. PT and T participants (n=47) from the neuropsychological cohort were recruited for the MRI study at age 20 years. In total, 17 Ts and 19 PTs, all with complete neuropsychological data at ages 16 and 18, met data quality criteria (described in Methods above) and were included in the imaging portion of the study. Demographic Characteristics Demographic data for the PTs and Ts for the 8, 12, 16, 18 and 20-year visits are shown in Table 1. The PTs and Ts included in all age cohorts were similar in gender makeup, race, and maternal education level. At ages 8, 12, and 16, there was a statistically significant difference in age between PTs and Ts at time of neuropsychological testing, likely due to consistent recruitment efforts for PTs for each visit around the time of their birthday, whereas Ts were recruited at any point during that year and therefore demonstrated increased age spread. Although this age difference may be clinically significant at age 8, it likely becomes clinically insignificant as the participants aged. There was no significant difference in the age at scan for PTs and Ts included in the imaged sub-cohort.