Alterations in cortical and subcortical neuroanatomy and associations with behavior in females with fragile X syndrome

Abstract

Aim:

To address substantial gaps in the literature on neuroanatomical variations in females with fragile X syndrome (FXS).

Method:

Surface-based modeling techniques were applied to the magnetic resonance imaging scans of 45 females with FXS (mean age = 10 years 9 months, range 6 years–16 years 4 months, SD = 2 years 9 months) and 33 age-matched and developmentally matched females without FXS to elucidate differences in cortical gray matter volume, surface area, and thickness. Gray matter volumes in subcortical regions were examined to ascertain differences in subcortical volume.

Results:

In females with FXS, cortical volume was greater bilaterally in the occipital pole and smaller in the right postcentral gyrus. Seven regions demonstrated lower surface area in participants with FXS, while cortical thickness was significantly greater over the posterior and medial surfaces in the group with FXS. Subcortical region of interest analyses demonstrated greater volume in the caudate nucleus, globus pallidus, and nucleus accumbens in the group with FXS. Global gray matter volume, pial thickness, and surface area were associated with behavioral outcomes in the group with FXS but not in the comparison group.

Interpretation:

Females with FXS demonstrated unique cortical and subcortical gray matter anatomy relative to a matched comparison group. These findings may be relevant to the pathogenesis of the FXS behavioral phenotype and provide insights into behavioral interventions targeted to this population.

Fragile X syndrome (FXS) is commonly cited as the leading monogenic cause of autism and intellectual disability. This heritable condition results from an expansion of approximately 200 (or more) CGG trinucleotide repeats in the FMR1 gene on the X chromosome and subsequent reduction in fragile X messenger ribonucleoprotein 1 protein (FMRP), a regulatory protein essential for brain function and development.¹ FMRP is a selective messenger RNA binding protein that regulates local protein synthesis. In its absence, its regulatory targets are overproduced, ultimately resulting in reduced synaptic strength and plasticity.²

Because FMR1 is located on the X chromosome, females (who have two X chromosomes) with FXS display a more heterogeneous phenotype than their male (XY) counterparts because of partial compensation from their second, unaffected X chromosome. In females, one of the two X chromosomes is randomly silenced early in embryonic development in each cell to ensure a balanced expression of X-linked traits relative to males.³ Because relatively few progenitor cells go on to form the brain, the variable ratio of affected versus unaffected active X chromosome results in a wide range of FMRP production and corresponding phenotypic expression across females with FXS.⁴

Females with FXS exhibit a range of cognitive, behavioral, and social-emotional difficulties.⁵ Specifically, individuals with FXS typically show relative strength in verbal abilities and relative weakness in attention, visual perception, visual construction, nonverbal reasoning, short-term memory, and executive function.^4,6,7 Behaviorally, individuals with FXS often experience difficulties with adaptive behavior, including social and communication skills. A previous study conducted by our group indicated that communication skills in particular are increasingly challenging for females with FXS as they age.⁸ Furthermore, FXS is associated with increased risk of autism, with as many as 60% of males and 14% of females with FXS meeting the diagnostic criteria.^9,10 Higher levels of autism symptomology are associated with poorer independent living skills in adults with FXS, an association that does not appear to exist for those with similar levels of idiopathic developmental disability.¹¹ Furthermore, individuals with FXS are at a significantly elevated risk for social-emotional challenges, including depression and anxiety. One study found that 56% and 22% of females with FXS were diagnosed with an anxiety or depressive disorder respectively.¹² While most available data use either male or mixed-sex groups, previous results published by our group using the current cohort found that females with FXS over 10 years 6 months of age experience significantly higher levels of depression, social avoidance, and withdrawal than their younger counterparts, suggesting that the severity of these characteristics may vary over time.¹³

Because FMRP has a role in protein regulation in the brain, there is substantial clinical interest in neuroimaging studies to characterize structural differences in FXS. In particular, the caudate nucleus, cerebellar vermis, hippocampus, amygdala, and fusiform gyrus exhibit significant volume differences in individuals with FXS compared to unaffected individuals.^14,15 These differences in regional brain volumes seem to persist throughout childhood, adolescence, and adulthood.^16–18

Because of the lower number of research studies focused on females with FXS, most investigations of brain differences in this population focus exclusively on males or on mixed-sex groups. One such study suggested that females with FXS have less total gray matter volume but greater total white matter volume compared to their male counterparts.¹⁶ A previous report from the current cohort demonstrated significantly increased gray matter volume in bilateral parieto-occipital regions and a more posterior right parieto-occipital region in addition to significantly reduced gray matter volume in the bilateral gyrus rectus compared to individuals without FXS. Regional volumes, in turn, were significantly associated with the key behavioral characteristics of FXS.¹⁹

This previous analysis conducted by our group relied on traditional voxel-based morphometry to quantify gray matter volume in cortical regions.¹⁹ Unlike newer surface-based modeling, voxel-based morphometry is not sensitive to the critical morphological characteristics that include gray matter volume, that is, thickness and surface area. These subcomponents of gray matter follow distinct neurodevelopmental trajectories,^20,21 have differential genetic influences,^22,23 and exhibit unique patterns of association with cognitive, socio-emotional, and behavioral outcomes in neurotypical individuals.^24,25 Thus, surface-based modeling incorporating both the thickness and surface area components of volume may provide additional insight into the distinctive neurodevelopmental profile of females with FXS, particularly considering the role of FMRP in synaptic pruning.⁶ In this study, we used these surface-based modeling approaches to examine cortical gray matter morphology in addition to region of interest (ROI) analyses to assess differences in subcortical and cortical ROIs.

The current study was designed to reduce the sex gap in understanding FXS and begin to elucidate brain mechanisms underlying the symptomology commonly seen in females with FXS, a population historically underrepresented in the literature.⁵ Due to the wide range of phenotypic expression and higher potential for successful outcomes relative to males with this condition, focusing research on females with FXS holds particular promise for improving our understanding of when and how to intervene to help affected females achieve their maximum potential.

METHOD

Participants

The group with FXS consisted of 45 females with a diagnosis of FXS confirmed by genetic testing demonstrating more than 200 CGG repeats accompanied by hypermethylation of FMR1 (mean age = 10 years 9 months, range 6 years–16 years 4 months, SD = 2 years 9 months). Participants in the group with FXS were recruited through the National Fragile X Foundation, support groups serving individuals with FXS, and social media and website announcements.

The comparison group consisted of 33 females identified as being at a similar risk for anxiety, avoidance, and arousal as females with FXS, including those with other genetic conditions such as KBG syndrome (n = 1), mosaic Down syndrome (n = 1), trisomy X (n = 3), and 7q11.23 duplication (n = 1); parent-reported diagnoses of autism spectrum disorder, dyslexia, or attention-deficit/hyperactivity disorder; and special educational needs. The comparison group was matched for sex, age, and adaptive behavior (Vineland Adaptive Behavior Scales, Third Edition [Vineland-3] Adaptive Behavior Composite) and were recruited through schools, organizations, and social media outlets targeting the desired population within the local area of the study site (mean age = 10 years 9 months, range 6 years 8 months–15 years 3 months, SD = 2 years 6 months).

Participants in either group were excluded from the study for very preterm birth (<32 weeks), magnetic resonance imaging (MRI) contraindications, sensory deficits that would preclude them from completing the study measures, or a diagnosis of a current or past seizure disorder, psychosis, bipolar disorder, or head trauma with loss of consciousness. Study procedures were conducted in accordance with the latest guidelines of the Declaration of Helsinki and approved by the Stanford Institutional Review Board for human research. Written informed consent was obtained from all caregivers and assent was obtained from each participant.

MRI preparation and completion

MRI scans were completed without sedation through behavioral training methods established for MRI procedures in young children as described previously by our group.¹⁹ MRI acquisition was performed on a SIGNA Premier 3T whole-body MRI system using a standard 48-channel head coil (GE Healthcare, Milwaukee, WI, USA). Axial T1 images of the brain were acquired using a three-dimensional magnetization-prepared, rapid-gradient, echo pulse sequence with the following parameters: repetition time = 1985ms; echo time = 2.8ms; inversion time = 900ms; flip angle = 8 degrees; slice thickness = 1.2mm; field of view = 24.0cm; acquisition matrix = 240 × 240; voxel size = 1.2 × 1.0 × 1.0mm; duration = 4:22 minutes. Scans were visually inspected at acquisition and repeated if significant motion artifacts were observed.

The complete study cohort included 102 participants. Seven scans could not be completed because of high participant anxiety precluding completion of the preparation protocol (five females with FXS, two females in the comparison group; mean age = 7 years 5 months). Nine scans could not be completed because the participant had braces or similar contraindications for MRI (three females with FXS and six females in the comparison group, mean age = 12 years 1 month). Eight scans were completed at the time of the study visit but excluded because of excessive motion that precluded analysis using the FreeSurfer software (http://surfer.nmr.mgh.harvard.edu) (three females with FXS, five females in the comparison group, mean age = 9 years 7 months). The resulting analytical cohort with usable MRI data for these analyses included 45 females with FXS and 33 females in the comparison group. Matching characteristics on the Vineland-3 Adaptive Behavior Composite did not differ significantly between the full cohort and the subset with usable MRI data (p = 0.57).

Cognitive and behavioral testing

Parents completed the following assessments to capture the behavior commonly affected by FXS: the Social Responsiveness Scale, Second Edition (SRS-2) to report on social function;²⁶ the Anxiety Depression and Mood Scales (ADAMS) to report on mood symptoms;²⁷ and the Vineland-3 to report on day-to-day functioning.²⁸

Each participant completed the following assessments: the Autism Diagnostic Observation Scale, Second Edition with a research-reliable examiner to assess the symptoms of autism; the Differential Abilities Scales, Second Edition (DAS-2) to assess cognitive abilities;²⁹ and the Kauffman Test of Educational Achievement, Third Edition (KTEA-3) to assess academic performance.³⁰

Image preprocessing

FreeSurfer v6.0 was used to estimate cortical thickness.^31–33 Briefly, processing streams included removing non-brain tissue, intensity normalization, segmenting gray and white matter, and aligning each image volume to a standardized space. To estimate cortical gray matter thickness, a deformable surface algorithm was applied to the segmented images to extract the pial and gray and white cortical surfaces.³¹ All cortical surfaces were visually inspected by study staff blinded to group membership; manual corrections were performed by blinded raters, where appropriate, according to previously established procedures.³⁴ This additional quality control step ensured the accuracy of gray and white matter segmentation, the exclusion of scalp and other non-brain tissue, and the inclusion of brain tissue.

Analyses

Cortical gray matter

Statistical analyses were conducted at each vertex to assess group differences in cortical gray matter volume, thickness, and surface area. Specifically, a general linear model was fitted at each vertex, with the structural measure (volume, thickness, surface area) as the dependent variable and diagnostic group (FXS, comparison) as the independent variable. Age was included as a covariate of non-interest for all three analyses, centered to the sample mean. Analyses of volume and surface area further controlled for total brain volume, again centered to the sample mean. Additionally, analyses of cortical thickness were conducted with and without the covariate of weighted mean thickness to elucidate regions with disproportionately increased thickness when accounting for overall cortical thickness. In all analyses, correction for multiple comparisons in the resulting statistical maps was conducted using a two-tailed threshold of p < 0.05 and the Monte Carlo simulation toolbox provided in FreeSurfer. This approach, which is based on the methods outlined by Hagler et al.,³⁵ estimates the probability of forming a maximum cluster of that size or larger during the simulation under the null hypothesis that results in a cluster-wise probability. We report cluster-wise probability values from vertex-wise analyses in the Results section of the article.

Subcortical gray matter

Subcortical ROIs were determined based on previous analyses of brain volume differences in individuals with FXS.^14–17 These ROIs were defined using the Desikan atlas and classified into subcortical regions.³⁶ A two-way multivariate analysis of covariance was conducted, including total brain volume and age as covariates. All assumptions were met to justify using multivariate analysis of covariance for the analyses. Planned follow-up tests were conducted to determine which specific regions drove the overall difference between the two groups.

Exploratory correlations

Exploratory associations between brain and behavioral metrics were conducted for the group with FXS and the comparison group separately. All assumptions were met to justify the use of Pearson’s rank correlation coefficients. To reduce the number of tests, correlation analyses were limited to overall measures of cortical gray matter volume, surface area, and thickness, as well as subcortical brain areas that were significantly different between the two groups in the primary analyses. Behavioral measures included standardized scores from the DAS-2, KTEA-3, and SRS-2 in addition to scores for the ADAMS Social Avoidance and Depressed Mood subdomains and the Autism Diagnostic Observation Scale, Second Edition total score.

RESULTS

Cortical gray matter

Group differences in cortical volume, thickness, and surface area are presented in Table S1 and Figures 1 to 4. Vertex-based analyses of cortical gray matter volume indicated a larger volume in the group with FXS relative to the comparison group in two clusters bilaterally centered in the posterior occipital pole. The group with FXS demonstrated reduced volume relative to the comparison group in two clusters: one encompassing the insula, superior temporal gyrus, and inferior frontal cortex; and another in a band extending from the precentral to the postcentral gyrus.

Figure 1:

Whole-brain cortical gray matter volume differences between the two groups, with age and total brain volume included as covariates of non-interest. Abbreviation: FXS, fragile X syndrome.

Figure 4:

Whole-brain gray matter thickness differences between the two groups, covarying for weighted mean thickness and age. Abbreviation: FXS, fragile X syndrome.

Vertex-based analyses of pial surface area indicated smaller size in the group with FXS relative to the comparison group in seven regions. On the left, these regions included clusters encompassing the paracentral, precentral, precuneus, and lateral orbitofrontal areas. On the right, these clusters encompassed the postcentral, precuneus, and middle temporal areas. No clusters of increased surface area were observed in the group with FXS relative to the comparison group.

Vertex-based analyses of cortical thickness indicated larger thickness over much of the posterior and medial surfaces in the group with FXS relative to the comparison group. After controlling for weighted mean thickness and age, the left cuneus, middle temporal region, and right precentral region remained significantly larger in the group with FXS relative to the comparison group.

Subcortical gray matter

There was a significant main effect of group in the analysis of subcortical gray matter (Table 1; F = 7.823, p < 0.001). Planned follow-up tests revealed that the volume of the caudate nucleus and nucleus accumbens was significantly larger in the group with FXS relative to the comparison group (p < 0.001), as was the globus pallidus (p < 0.05). No between-group difference was observed in gray matter volume of the thalamus, putamen, hippocampus, or amygdala.

Table 1:

Demographic characteristics and global cortical metrics in the fragile X syndrome and comparison groups

Characteristic	Group with FXS, (n = 45)	Comparison group, (n = 33)	p
Age, years:months, mean (SD)	10:9 (2:9)	10:9 (2:6)	0.92
Ethnicity, n			0.54
Asian	2	6
Black	1	1
White	40	16
Multiracial	1	9
Not reported	1	1
Ethnic group, n			<0.001
Hispanic or Latino	4	13
Not Hispanic or Latino	40	20
Not reported	1	0
Adjusted SES, mean (SD)	1.27 (0.72)	1.11 (0.54)	0.28
Parental education, n			0.50
Partial high school	0	1
High school graduate	1	0
Partial college	7	3
College graduate	19	14
Graduate degree	15	15
Not reported	3	0
Child assessment, mean (SD)
Vineland-3 ABC	79.22 (10.92)	76.48 (10.50)	0.27
DAS-2 verbal abilities	81.00 (16.49)	87.60 (18.53)	0.11
ADOS-2 classification, n			0.16
Non-spectrum	26	20
Spectrum	5	5
Autism	14	8
Global brain measures, mean (SD)
Mean weighted thickness	2.86 (0.13)	2.76 (0.11)	<0.001
Total pial surface area	225 092 (17 434)	214 316 (19 471)	0.014
Total gray matter volume	578 211 (52 812)	533 356 (53 479)	<0.001

Abbreviations: ABC, Adaptive Behavior Composite; ADOS-2, Autism Diagnostic Observation Scale, Second Edition; DAS-2, Differential Abilities Scales, Second Edition; FXS, fragile X syndrome; SES, socioeconomic status.

Exploratory brain–behavior associations

The Pearson’s rank correlation coefficients between each brain measure and the scores on the DAS-2 General Conceptual Ability, KTEA-3 Brief, Autism Diagnostic Observation Scale, Second Edition Total, SRS-2 Total, and ADAMS Depressed Mood and Social Avoidance measures in the group with FXS are shown in Table 2. No correlations were observed between behavior and the volume of the caudate nucleus or nucleus accumbens. Mean thickness was positively associated with the DAS-2 General Conceptual Ability and KTEA-3 Brief Composite, and negatively associated with the ADAMS Depressed Mood. The total pial surface area was positively associated with the DAS-2 General Conceptual Ability and KTEA-3 Brief, and negatively associated with the SRS-2 Total, ADAMS Depressed Mood, and ADAMS Social Avoidance. Total gray matter volume was positively associated with the DAS-2 General Conceptual Ability and KTEA-3 Brief, and negatively associated with the SRS-2 Total, ADAMS Depressed Mood, and ADAMS Social Avoidance.

Table 2:

Results from the planned follow-up tests of significant subcortical multivariate analysis of covariance

Brain subcortical areas	Group with FXS (n = 45)	Comparison group (n = 33)	F	p
Thalamus	15 305 (1169)	14 464 (1430)	0.44	0.509
Caudate nucleus	8564 (1032)	6974 (1095)	23.16	<0.001
Putamen	10 678 (1109)	9892 (1284)	0.45	0.503
Globus pallidus	3503 (519)	3049 (543)	5.08	0.027
Hippocampus	7428 (700)	7086 (892)	1.62	0.207
Amygdala	2738 (526)	2560 (466)	0.29	0.595
Nucleus accumbens	2738 (526)	1332 (221)	15.27	<0.001

Values for the FXS and comparison groups are shown as the uncorrected mean (SD). Effect sizes and p-values were calculated from the multivariate analysis of covariance. Abbreviation: FXS, fragile X syndrome.

In the comparison group, only the association between the ADAMS Social Avoidance and nucleus accumbens was significant (r = −0.44, p ≤ 0.05); the groups showed significantly different Fisher r to z transformations for the association between pial surface area and total gray matter volume with the SRS-2 Total score (p = 0.005 and 0.04), ADAMS Depressed Mood (p = 0.03 and p = 0.008), and ADAMS Social Avoidance (p = 0.048 and p = 0.050) (Table 3).

Table 3:

Exploratory associations in each group between significant subcortical regions of interest and global brain metrics with the DAS-2 GCA, KTEA-3 Brief, ADOS-2 Total, SRS-2 Total, and ADAMS Depressed and Social Avoidance measures

ROI/global brain metric	DAS-2 GCA		KTEA-3 Brief		ADOS-2 Total		SRS-2 Total		ADAMS Depressed		ADAMS Social Avoidance
	FXS	Comparison	FXS	Comparison	FXS	Comparison	FXS	Comparison	FXS	Comparison	FXS	Comparison
Caudate nucleus	−0.10	−0.02	−0.10	0.01	0.15	−0.03	0.02	0.12	0.08	0.16	−0.07	0.14
Nucleus accumbens	0.18	−0.22	0.25	−0.05	−0.23	−0.21	−0.16	−0.22	−0.25	−0.31	−0.16	−0.44*
Globus pallidus	−0.30*	0.06	−0.13	0.09	0.25	−0.05	−0.25	−0.10	0.09	−0.03	−0.14	−0.10
Mean thickness	0.37*	0.28	0.36*	0.18	−0.41**	−0.27	−0.01	−0.10	−0.40**	−0.08	−0.22	−0.08
Pial surface area	0.34*	0.10	0.36*	0.07	−0.09	0.18	−0.42 **	0.22	−0.32 *	0.18	−0.39 **	0.06
Gray matter volume	0.42**	0.16	0.47**	0.14	−0.26	0.04	−0.34 *	0.13	−0.44 **	0.16	−0.41 **	0.03

The values in bold indicate significant Fisher r to z differences in correlation strength between the two groups.

^*p < 0.05

^**p < 0.01

^***p < 0.001.

Abbreviations: ADAMS, Anxiety Depression and Mood Scales; ADOS-2, Autism Diagnostic Observation Scale, Second Edition; DAS-2, Differential Abilities Scales, Second Edition; FXS, fragile X syndrome; GCA, General Conceptual Ability; KTEA-3, Kauffman Test of Educational Achievement, Third Edition; ROI, region of interest; SRS-2, Social Responsiveness Scale, Second Edition.

DISCUSSION

This study was conducted to examine FXS-specific alterations in cortical volume, surface area, and thickness, as well as differences in subcortical and cortical volume. Using advanced surface-based modeling, cortical gray matter volume was shown to be larger bilaterally in the occipital pole in females with FXS; females with FXS demonstrated smaller volume in the right postcentral gyrus. Examination of the cortical surface area revealed seven distinct regions with lower surface area in participants with FXS. Cortical thickness in females with FXS was significantly higher than in females in the comparison group over much of the posterior and medial surfaces. Complimentary subcortical ROI analyses demonstrated larger volume in the caudate nucleus, globus pallidus, and nucleus accumbens in the group with FXS relative to the comparison group. Finally, exploratory correlations demonstrated minimally significant associations in the comparison group, while mean thickness, total surface area, and total gray matter volume were significantly associated with cognitive, academic, social, and mood outcomes in the group with FXS. Together, these data provide further evidence for a distinct neuroanatomical phenotype in school-age females with FXS and extend previous imaging findings by elucidating specific surface area and thickness differences that contribute to the overall cortical phenotype in this population.

The subcortical ROI findings in this study demonstrate elevated volumes in females with FXS in the caudate nucleus, globus pallidus, and nucleus accumbens. While an enlarged caudate nucleus has been well documented in research on FXS,^14–16 our study is the first to demonstrate that the globus pallidus and nucleus accumbens are enlarged in school-age females with FXS. Both regions are components of the broader basal ganglia, which is a critical difference in neuroanatomy in individuals with FXS.⁶ A larger volume in these two substructures may represent a neuroanatomical signature unique to school-age females with FXS. Surprisingly, no significant differences were observed in either the hippocampus or the amygdala, both of which have previously been demonstrated to be larger in individuals with FXS.^37–39 This difference from previously reported volume differences may represent an important distinguishing feature in females with FXS or may be associated with the presence of partially intact FMRP production. Furthermore, the thalamus did not exhibit differences in the group with FXS relative to the comparison group, although previous investigations suggested that thalamus enlargement was unique to females with FXS.^16,39 This difference in study results may be due to the use of a typically developing comparison group in previous studies as opposed to a heterogeneous, developmentally matched group in the present study, or may be a consequence of the shift from manual to automatic regional segmentation in FreeSurfer. Exploratory correlations in our cohort with FXS revealed a significant association between the globus pallidus and overall cognitive abilities, but no other associations between these regions and outcomes of interest. This indicates that while differences in these subcortical regions may contribute to the phenotype in females with FXS, they are not sufficient individually to explain phenotypic variation in this population. As new, targeted treatments for FXS become more widely available, these regions may be important for the development of specific gene therapies for females with FXS in particular.⁴⁰

Surface-based modeling in the present study represents a unique contribution to the literature in elucidating the surface area and cortical thickness components associated with volume differences in this population. While all regions with a significantly different surface area were larger in the comparison group, all regions with significant thickness findings were thicker in the group with FXS, unless the weighted mean thickness was included in the model. Given that previous findings suggested that maximum surface area and minimum thickness facilitate brain activity and development, and that cortical thinning in adolescence is associated with higher IQ,^21,25 this specific combination of alterations may have an important role in the FXS phenotype. FMRP has a critical role in synaptic remodeling.^6,15,41 Therefore, increased thickness across widespread areas of the brain in individuals with FXS may be at least partially attributable to FMRP-associated reductions in synaptic pruning. As such, cortical thickness could be considered a potential robust brain biomarker for FXS and a treatment end point in future treatment trials. Finally, while overall cortical thickness, surface area, and gray matter volume demonstrated no significant associations with outcomes in the comparison group, strong associations were demonstrated in the group with FXS. Furthermore, the strength of the correlations between the two groups was significantly different for measures of social skills, social anxiety, and depression, providing further evidence for brain–behavior associations unique to this population.

We acknowledge several limitations to the present study. While every effort was made to recruit a diverse population in both groups, the study population was predominately White and the parents of most participants had obtained at least a 4-year college degree, which may limit the generalizability of our findings. Additionally, the comparison group was recruited locally, while the group with FXS was recruited throughout North America, potentially affecting the findings. The comparison group consisted of a heterogeneous group of females with diverse risk factors for learning and behavioral challenges, while a typically developing group was not investigated to provide a baseline for brain differences. Furthermore, among females with FXS, the activation ratio is associated with the severity of the phenotype. In an effort to reduce participant burden in this vulnerable population, we did not conduct blood draws to measure the activation ratio, so these data are not available to compare them to the imaging findings. Future imaging studies in this population should consider investigating the association between the activation ratio and neuroanatomy. As another method to reduce participant burden, all participants completed the MRI without sedation. However, this resulted in several participants being excluded because of anxiety or motion artifacts. While the subset of participants with successful MRI scans do not differ from the complete cohort on the Vineland-3, they may differ on measures of anxiety that may impact exploratory associations. However, because more participants with FXS were excluded because of anxiety than in the comparison group, and the associations with the ADAMS subscales were only seen in the group with FXS, we believe that these exclusions would bias our findings toward the null hypothesis; inclusion of these participants would only demonstrate a greater signal. Finally, automated methods in FreeSurfer may result in noisier parcellation of subcortical regions than if manual methods were used; future studies should consider evaluating white matter differences in FXS in addition to gray matter neuroanatomy.

Overall, the present study provides a valuable contribution to FXS research by clarifying the cortical morphometry unique to this population, including not only volumetric differences but also the surface area and thickness components unique to individuals with FXS. Further longitudinal analyses that examine the effects of FXS on neuroanatomy over time, given the many neuroanatomical changes that occur through childhood and early adolescence, are warranted.

Supplementary Material

Supinfo

Table S1: Values for the FXS and comparison groups.

The following additional material may be found online:

Figure 2:

Whole-brain gray matter surface area differences between the two groups, with age and total brain volume included as covariates of non-interest. Abbreviation: FXS, fragile X syndrome.

Figure 3:

Whole-brain gray matter thickness differences between the two groups, with age included as a covariate of non-interest. Abbreviation: FXS, fragile X syndrome.

What this paper adds.

• Cortical volume was greater in the occipital pole of females with fragile X syndrome (FXS).

• Cortical surface area was smaller across a large portion of the brain in females with FXS.

• Cortical thickness was greater in the posterior and medial surfaces in females with FXS.

• The volumes of the caudate nucleus, nucleus accumbens, and globus pallidus were greater in females with FXS.

• Mean cortical thickness, total pial surface area, and total brain volume were associated with key phenotypic traits in females with FXS.

ABBREVIATIONS

ADAMS

Anxiety Depression and Mood Scales

DAS-2

Differential Abilities Scales, Second Edition

FMRP

fragile X messenger ribonucleoprotein 1 protein

FXS

fragile X syndrome

KTEA-3

Kauffman Test of Educational Achievement, Third Edition

ROI

region of interest

SRS-2

Social Responsiveness Scale, Second Edition

Vineland-3

Vineland Adaptive Behavior Scales, Third Edition