Genomic Imprinting Effects on Cognitive and Social Abilities in Prepubertal Girls with Turner Syndrome

Abstract

Context:

Recent evidence suggests that the cognitive and social manifestations associated with Turner syndrome (TS) might be influenced by epigenetic factors in the form of genomic imprinting. However, due to small and heterogeneous samples, inconsistent results have emerged from these studies.

Objective:

The objective of this prospective study was to establish the impact of genomic imprinting on neurocognitive abilities and social functioning in young girls with TS.

Design, Setting, and Participants:

An extensive battery of neuropsychological assessments was administered to 65 children with TS who had never been exposed to estrogen treatment, 24 of whom had an X-chromosome from paternal origin (X^pat) and 41 from maternal origin (X^mat).

Main Outcome Measures:

The Wechsler scales of intelligence, the Motor-Free Visual Spatial test-3, the Wide Range Assessment of Visual Motor Ability, and the attention/executive domain of the NEPSY were used to assess cognitive abilities. Social functioning was assessed with the Social Responsiveness Scale and the Behavior Assessment System for Children-2.

Results:

Results showed that although individuals with X^pat obtained lower scores than their counterparts with X^mat on most cognitive and social measures, only the Perceptual Reasoning Index of the intelligence scale yielded significant differences after correction for multiple comparisons.

Conclusion:

Overall, these results suggest that although some aspects of the neuropsychological profile of TS may be influenced by epigenetic factors, the sociocognitive phenotype associated with the disorder is not modulated by genomic imprinting.

Turner syndrome (TS) is a genetic disorder resulting from the complete or partial absence of one copy of the X-chromosome in females. In individuals with complete X-monosomy, the process of X-inactivation that takes place in typically developing females is bypassed, resulting in the exclusive expression of gene products from a single X-chromosome in each active cell, which is either paternally (X^pat) or maternally (X^mat) inherited. This particular expression pattern provides a unique opportunity to study the potential effect of genomic imprinting, a process by which some genes are preferentially activated depending on their parental origin, on cognition.

The potential impact of genomic imprinting of the X-chromosome has received much interest, especially given that imprinted genes on this chromosome may be expressed in a sex-specific manner, thereby contributing to certain sex-related phenotypes (1). So far, conflicting results have emerged from studies pertaining to imprinting effects on a small number of physical (2, 3) and cognitive (4, 5) characteristics of TS. In a seminal paper, Skuse and colleagues (4) reported that X^mat individuals presented significant impairments in social and executive functioning in comparison with their X^pat counterparts. Although this paper received much attention and sparked interest in studying the effects of genomic imprinting of the X-chromosome on cognition (4–8), these results have not been replicated. Furthermore, conclusions from subsequent studies in this area are limited by small sample sizes. Clarifying this issue is of great importance for clinical practitioners who treat individuals with TS, with respect to the identification of patients at potentially increased risk for behavioral dysfunction, and for the field of psychiatric/behavioral genetics in general.

To clarify the impact of genomic imprinting on cognition and social functioning, we conducted a prospective study using an extensive battery of neuropsychological tests and questionnaires to assess neurocognitive and social abilities in a cohort of 65 young girls with complete X-monosomy (41 X^mat, 24 X^pat) who have never been exposed to exogenous estrogen treatment. Results from our study, the largest to date to address this issue, suggest the presence of imprinting effects of the X-chromosome that affect visuoperceptual reasoning abilities in TS but without affecting sociocognitive functions. These results argue against a hypothesis of imprinted genes on the X-chromosome that affect social cognition.

Subjects and Methods

Subjects

Participants were recruited from national TS organizations in the United States and Canada, by referral from their treating pediatric endocrinologists, and through the Stanford Center for Interdisciplinary Brain Sciences Research website (http://cibsr.stanford.edu). Written informed consent was obtained from all parents, and verbal assent was obtained from children. Study procedures and protocols were approved by the Stanford University research compliance office.

Participants were included in the present study if they met the following criteria: 1) aged between 3 and 12 yr old; 2) presenting a cytogenetic status of nonmosaic 45,X; 3) willingness and availability of at least one biological parent to provide a peripheral blood sample and demographic information and take part in an IQ test; and 4) not presenting any chronic condition other than those associated with TS. X-monosomy was established through standard karyotype analyses of at least 20 cells, which allows exclusion of 11% mosaicism or greater with 0.90 confidence (9). Sixty-five participants with TS met these criteria (range 3–12 yr; mean age 8.16 ± 2.46 yr), of which 24 were X^pat (mean age 8.14 ± 2.81 yr) and 41 were X^mat (mean age 8.17 ± 2.27 yr). Parental origin of the X-chromosome was determined by comparison of amplification patterns of four polymorphic markers located exclusively on the X-chromosome (DXS6807 at 14 cM, DXS993 at 42 cM, DXS1106 at 67cM, and the androgen receptor at 90 cM) and one marker in the pseudoautosomal region (amelogenin) between the proband and parents. Hormonal treatment history of all TS subjects was obtained through self-report or medical files; 49 participants (30 X^mat) had taken GH at the time of testing, and no participant reported adhering to an estrogen replacement therapy. Detailed participant characteristics are presented in Table 1.

Table 1.

Participant characteristics and neuropsychological test results

	Xmat (n = 41)	Xpat (n = 24)	P value
Demographics
Age (yr)	8.17 (2.27)	8.14 (2.81)	0.965
GH treatment (n)	30/41	19/24	0.588
Parent's FSIQ	113.32 (11.13)	112.55 (11.13)	0.797
Family income	6.64 (1.84)	6.89 (2.51)	0.665
Neuropsychology
WISC/WPPSI
FSIQ	95.00 (14.75)	89.00 (12.73)	0.101
VCI	105.98 (12.67)	99.35 (13.86)	0.039
PRI	95.93 (15.77)	84.38 (13.11)	0.004a
WMI	84.06 (13.67)	88.89 (13.15)	0.225
PSI	89.34 (13.42)	89.52 (13.71)	0.961
NEPSY attention/executive functions	91.68 (14.50)	87.70 (15.89)	0.315
MVPT-3	83.42 (21.75)	76.42 (16.39)	0.229
WRAVMA
Composite	77.94 (14.98)	80.10 (17.41)	0.641
Drawing	89.71 (13.93)	92.95 (18.12)	0.469
Matching	86.81 (13.57)	81.57 (13.24)	0.174
Pegboard	75.68 (17.05)	82.29 (16.26)	0.169

Demographics and summary of the neuropsychological assessment for both groups (X^mat and X^mat). Means are presented with sd in parentheses. One-way ANOVA were used for all comparisons, except for GH treatment, where a χ² test was used. Family income was assessed with a self-report questionnaire using a 10-point Likert scale. Significance level was established at α = 0.05 for demographics and at α = 0.004 (Bonferroni corrected) for neuropsychological tests. PSI, Processing Speed Index; WMI, Working Memory Index.

^aSignificant comparisons.

Neuropsychological assessment

Neuropsychological assessments were conducted by a research psychologist over two consecutive days. Depending on participants' age, the Wechsler Preschool and Primary Scale of Intelligence Third Edition or the Wechsler Intelligence Scale for Children Fourth Edition was used to measure children's Full Scale Intelligence Quotient (FSIQ), Verbal Comprehension Index/Verbal IQ (VCI), Perceptual Reasoning Index/Performance IQ (PRI), and Processing Speed Index/Quotient as well as Working Memory Index for participants who took the Wechsler Intelligence Scale for Children Fourth Edition (19 X^pat; 31 X^mat). Parent's IQ was measured using the Wechsler Abbreviated Scale of Intelligence. The Wide Range Assessment of Visual Motor Abilities (WRAVMA), designed to assess visuomotor skills in children and adolescent from 3–17 yr old, was used to assess fine-motor, visual-spatial, and visual-motor abilities. Reliability measures of the three subtests of the WRAVMA show internal consistency coefficients exceeding 0.90 and test-retest reliability coefficients ranging from 0.81–0.91 (10). In addition, the Motor-Free Visual Perception Test Third Edition (MVPT-3) was used to measure visual perception without the contribution from the participant's motor skills, an area of known difficulty in TS. The MVPT-3 is a 65-item test using black-and-white stimuli and response choices; it presents good internal consistency (∼0.80) and discriminative validity (11). The Attention-Executive Functions Domain score of the NEPSY (12) was used to quantify executive functions, including the ability to maintain attention toward auditory stimuli, shift set, planning, self-regulating, monitoring, and problem solving.

The Social Responsiveness Scale (SRS) (13) and the Behavior Assessment System for Children Second Edition (BASC-2) (14) were used to measure social functioning. The SRS is a parental report designed for children 4–18 yr in age and gives a total score that sums scores obtained on five subscales: 1) social awareness, 2) social cognition, 3) social communication, 4) social motivation, and 5) autistic mannerisms. The SRS is a quantitative measure of autistic traits that has excellent psychometric properties and correlates strongly with established diagnostic tools of autism spectrum disorders.

The BASC-2 is a parent report questionnaire that assesses emotional and behavioral functioning in individuals aged 2–18 yr. It comprises nine clinical scales (hyperactivity, aggression, conduct problems, anxiety, depression, somatization, atypicality, withdrawal, and attention problems), five adaptive scales (adaptability, social skills, leadership, activities of daily living, and functional communication), and four composite scores (externalize problems, internalize problems, behavioral symptoms, and adaptive skills); this instrument has been shown to have good to excellent reliability and validity (14).

All tests were administered and scored following standard procedures. Results of the tests and questionnaires were scale-scored and age-normed, except for the SRS for which no age-norm exists. To address this issue, age was used as covariate in the statistical analyses using SRS as an outcome measure.

Statistical analysis

Statistical analyses were performed using SPSS version 19.0 software. Group differences on all measures were assessed using one-way ANOVA for age-normed assessments. For the SRS subscales, analyses of covariance (ANCOVA) using age as covariate were performed. In addition, social measures (SRS and BASC-2) were analyzed using univariate ANCOVA with FSIQ as covariate to account for the influence of global intellectual abilities on social functioning. Significance levels were established using Bonferroni corrections to adjust for multiple testing: neuropsychological tests (α = 0.004) and social questionnaires (α = 0.002).

Results

Demographic features (age, gender, parental IQ, and socioeconomic status) and medical history regarding GH were not significantly different across groups (all P ≥ 0.665) (Table 1). Analyses of neuropsychological tests showed that the groups differed significantly on the PRI, where girls with X^mat scored significantly higher than individuals with X^pat (P = 0.004). Although no significant difference was observed on any of the other neurocognitive measures, a trend in the same direction was noted on the VCI (P = 0.039) (Table 1). Analyses conducted with measures assessing social cognition, the SRS and BASC-2, revealed no significant differences between groups with or without FSIQ as a covariate (Table 2).

Table 2.

Sociocognitive measures

Sociocognitive measures	Xmat (n = 41)	Xpat (n = 24)	P value	ANCOVA (FSIQ)
				Model P value	X-origin P value
BASC-2
Clinical scales
Hyperactivity	56.45 (16.19)	59.42 (12.60)	0.448	0.189
Aggression	47.26 (11.42)	51.00 (8.42)	0.172	0.375
Conduct problems	45.38 (11.28)	51.68 (7.06)	0.033	0.081
Anxiety	53.53 (13.96)	57.17 (13.63)	0.317	0.324
Depression	48.74 (12.97)	54.88 (12.29)	0.069	0.016	0.157
Somatization	50.39 (13.69)	52.54 (10.42)	0.514	0.801
Atypicality	53.79 (15.12)	61.46 (15.99)	0.062	0.006	0.156
Withdrawal	51.34 (15.36)	52.63 (12.54)	0.733	0.143
Attention problems	53.50 (14.70)	59.46 (8.03)	0.074	0.003	0.197
Adaptive scales
Adaptability	46.82 (12.57)	45.63 (9.35)	0.691	0.201
Social skills	54.42 (12.55)	54.00 (10.09)	0.890	0.990
Leadership	47.53 (13.76)	46.89 (8.78)	0.857	0.016	0.849
Activities of daily living	44.45 (13.47)	43.75 (10.28)	0.829	0.243
Functional communication	45.55 (12.50)	44.83 (11.88)	0.819	0.016	0.734
Composite scores
Externalize problems	50.13 (12.45)	54.63 (9.39)	0.135	0.145
Internalize problems	51.39 (12.96)	56.00 (10.89)	0.153	0.118
Behavioral symptoms	52.76 (13.51)	58.38 (10.33)	0.088	0.005	0.220
Adaptive skills	47.92 (12.81)	46.58 (9.92)	0.661	0.045	0.966
SRS
Total SRS	62.08 (15.77)	66.33 (12.94)	0.554	0.001a	0.724
Autistic mannerism	66.38 (16.49)	71.26 (15.11)	0.519	0.008	0.591
Social awareness	58.69 (13.51)	63.11 (9.92)	0.316	0.010	0.429
Social communication	59.61 (15.82)	63.65 (13.39)	0.576	0.004	0.722
Social cognition	61.93 (14.53)	65.57 (14.02)	0.632	0.001a	0.826
Social motivation	55.17 (11.86)	57.04 (12.32)	0.799	0.003	0.894

Summary of the social questionnaires results for both groups (X^mat, X^mat). Significance levels for each comparison are shown (P values). For the SRS, P values indicate the significance of the model for the univariate ANCOVA using age as covariate. The last two columns show P values for univariate models (ANCOVA) using FSIQ as covariate (both age and FSIQ for SRS). When the overall model's P value is <0.05, significance of the X-origin factor is indicated. Significance threshold was established using Bonferroni corrections (α = 0.002).

^aSignificant comparisons.

Discussion

Results of the current study, the largest to date to investigate social and neurocognitive functions in a homogeneous cohort of TS with regard to estrogen treatment, argue convincingly against the existence of a genomic imprinting effect of the X-chromosome that influences social cognition in these young girls. On the other hand, our results suggest the presence of genomic imprinting effects on visuoperceptual reasoning abilities, an area of known difficulty in TS.

The only significant difference found in this comprehensive neurocognitive assessment was in favor of X^mat on the PRI. The PRI probes organization and reasoning abilities using material that is visually presented. In that sense, it relies both on executive functions and visuospatial processing. Our results are contradictory to previous reports of superior executive functions in X^pat (4) but are in general agreement with those of Loesch et al. (5) who reported that X^pat were more severely impaired than X^mat on a number of measures of the Wechsler scales, including the VCI and FSIQ. Although differences on these scales did not reach significance in our sample, a trend in the same direction was observed (VCI, P = 0.039; FSIQ, P = 0.101). It is interesting that other measures tapping individual executive functions (NEPSY attention/executive domain) and visuospatial skills (MVPT-3, WRAVMA) did not differ between groups. It is plausible that X^pat individuals may be more vulnerable than X^mat when it comes to recruiting additional cognitive resources needed to face challenging visuospatial problems, a phenomenon that has been previously documented in TS (15, 16).

Contrary to our expectations, individuals with X^pat and X^mat did not differ significantly on any of the 24 measures/scales pertaining to social cognition. This observation is strengthened by the fact that analyses taking into account global intellectual functioning also did not demonstrate the presence of imprinting effects on social aptitudes. These results are in apparent contradiction with the original study of Skuse and colleagues (4), who found that X^pat individuals were less severely affected than their counterparts with X^mat in the social domain. However, it must be noted that in their study, Skuse et al. (4) included a cohort of older individuals with TS with much broader age range (6–18 yr). Thus, one explanation reconciling the present findings with that study may be that puberty drives significant changes in brain functions of individuals with TS and potentially interacts with imprinted genes to result in different cognitive-behavioral phenotypes across the developmental lifespan. However, investigation of this hypothesis requires a longitudinal study that spans the pubertal period.

In summary, our results indicate no effect of genomic imprinting on standardized measures of social cognition. However, the present results do suggest the existence of a genomic imprinting effect on the X-chromosome that affects visual-perceptual reasoning in individuals with TS. Although this study is one of the largest conducted on the effect of genomic imprinting on cognition, the sample is still relatively small and may not be sufficient to detect subtle differences between groups. Future studies are needed to confirm this finding and, if replicated, establish a specific genetic foundation for this phenomenon. Recent animal models have shown the existence of preferential gene expression in a sex-specific manner in the mammalian brain (17); however, such genes have yet to be identified in humans. Results from a previous study with a small sample also suggest that individuals with X^pat and X^mat differ with regard to brain morphology (18). Thus, future investigation using neuroimaging to study brain structure, function, and development might be a more sensitive method for establishing neural phenotypes associated with genomic imprinting in TS. In the absence of compelling evidence from future research, researchers and clinicians should be vigilant in interpreting reports of potential X-chromosome-related epigenetic effects on behavior, and be especially critical when assessing the clinical relevance of genomic imprinting for their patients.