ADHD and ASD Symptoms in Young Males with Fragile X Syndrome: Associations with Early Trajectories of Inhibitory Control

Abstract

Inhibitory control (IC), the ability to suppress inappropriate responses, emerges late in the first year of life and improves across typical development, concurrent with brain maturation. The development of IC is critical to various social-emotional and behavioral functions, with IC difficulties being linked to numerous neurodevelopmental disorders, including attention-deficit/hyperactivity disorder (ADHD) and autism spectrum disorder (ASD). Fragile X syndrome (FXS) is a single-gene disorder characterized by IC difficulties, and elevated rates of ADHD and ASD, making it a useful model for understanding the early development and consequences of IC. In this longitudinal study, we characterized IC trajectories across multiple time points between 16 and 71 months of age in young males with FXS (n=79) relative to neurotypical (NT) controls (n=49). To explore the association between behavioral outcomes and IC, we identified a subsample of 50 children with longitudinal IC data and an outcome assessment for ADHD and ASD symptoms at age 5 (FXS: n=26, NT: n=24). Results indicated that, compared to their NT peers, young males with FXS exhibit differences in IC as early as 24 months, with group differences increasing through age 5. Additionally, we determined that lower IC levels at 24 months were associated with later ADHD symptoms and a decreasing slope in IC over time was associated with later ASD symptoms in male children with FXS. These findings help refine early developmental phenotypes of FXS and highlight IC as a potential target for early detection and intervention of ASD and ADHD symptoms in male children with FXS.

Inhibitory Control in Development

Inhibitory control (IC) is the voluntary ability to withhold inappropriate responses and behaviors; it is a foundational component of executive functioning (EF), a multi-faceted construct which underlies self-regulation and goal-directed behavior (Diamond, 2013; Friedman et al., 2008; Garon et al., 2008; Miyake et al., 2000; Zelazo et al., 2008; Zelazo & Muller, 2011). The emergence of IC in early childhood can be well conceptualized according to neuroconstructivist and broader developmental theories (Hodapp et al., 1990; Karmiloff-Smith, 2006). The neuroconstructivist theory stipulates that processes at various levels, including genetic, neural, cognitive, and behavioral, exert bidirectional influences on each other across development. Indeed, increasing specialization and efficiency of information processing and behavioral control across the first few years of life is simultaneously supported by, and reinforces, the rapid development of the prefrontal cortex and increasing connectivity among neural structures (Buss & Spencer, 2014; Kochanska et al., 2000; Rothbart & Posner, 2001). Further, the development of IC and underlying neural structures enables the development of higher-order cognitive functions and more sophisticated goal-directed activity and emotion regulation (Zelazo & Cunningham, 2007).

In this study, we examine the trajectory and consequences of early IC development in male children with Fragile X syndrome (FXS), a single-gene disorder characterized by difficulties in IC and high prevalence of diagnostic comorbidities including attention-deficit/hyperactivity disorder (ADHD) and autism spectrum disorder (ASD). Examining IC as a potential early risk marker may improve our understanding of the multifinality of behavioral outcomes, such as ADHD and ASD symptoms, in individuals with FXS and may contribute to identifying mechanistic underpinnings of these behaviorally-defined disorders (Friedman et al., 2008).

Inhibitory control emerges late in the first year of life, improves steadily, and becomes increasingly differentiated from other behavioral abilities over time in neurotypical development (see Garon et al., 2008 for review; Diamond, 2013; Hongwanishkul et al., 2005; Rothbart et al., 2007). The expected improvement with age in regulatory abilities such as IC results in greater heterogeneity of these abilities across individuals, with rank-order stability, the maintenance of one’s position relative to the group over time, of IC between individuals across development being largely attributed to genetic factors (Gagne & Saudino, 2016; Putnam et al., 2008). Both task performance and parent ratings of IC show demonstrated improvement across early childhood (Petersen et al., 2016, Putnam et al., 2006). For example, Carlson (2005) examined performance of 2- to 6-year-olds on a battery of commonly used EF tests and found age-related improvements in IC task performance across this age range. Though there is no consensus on the age at which IC is fully mature (Petersen et al., 2016), findings by Macdonald et al. (2014) indicate that children commit fewer inhibition errors on an IC task, therefore successfully inhibiting a dominant response, and engage in more self-correction of errors, demonstrating greater awareness of errors, with age from ages 5 to 8 years. Moreover, maturation and differentiation of IC and higher-order executive functions continues through adolescence and into early adulthood (Constantinidis & Luna, 2019; Ordaz et al., 2013; Shing et al., 2010). Previous research has also revealed sex differences in IC, with males evidencing similar or lower parent ratings of IC and performing more poorly on IC tasks compared to their female peers (Eisenberg et al., 2005; Garon et al., 2016; Memisevic & Sinanovic, 2014; Thorell et al., 2004).

Inhibitory control contributes to the development of social-emotional skills and self-regulation (Kochanska et al., 2000). Indeed, greater IC in childhood has been linked to a number of prosocial behaviors and skills, including social competence, committed compliance, and theory of mind (Carlson et al., 2004; Dyson et al., 2012; Eisenberg et al., 2007; Eisenberg & Spinrad, 2004; Lengua, 2003; Liew et al., 2004; Rhoades et al., 2009; Rothbart & Bates, 2006; Spinrad et al., 2012). Greater IC also mitigates risk for the development of poor adjustment outcomes, with high levels of IC predicting decreased incidence of internalizing and externalizing behaviors (Kim-Spoon et al., 2019; Lengua, 2003; Muris & Ollendick, 2005; Rhoades et al., 2009). Furthermore, IC impairments have been associated with diminished self-regulatory skill development, poor emotion regulation, and the development of behavior problems (Brunsdon & Happé, 2014; Carlson & Wang, 2007; Diamond, 2013; Espy et al., 2011; Miyake & Friedman, 2012). Further research is needed to confirm whether IC is a useful marker and target for the early detection and prevention of specific behavioral phenotypes. Mapping the development of IC in early childhood is crucial to elucidating when deviant trajectories become apparent and thus identifying sensitive periods for intervention. Targeted intervention of IC in early childhood may have the cascading effects of altering a child’s developmental trajectories and therefore possibly minimizing achievement gaps or health inequities (Diamond, 2015).

Behavioral Outcomes Associated with IC: ADHD and ASD

ADHD.

An extensive body of research upholds that IC deficits underly the development of ADHD (Barkley, 1997). For example, DePauw and Mervielde (2011) found that, compared to NT children, children with ADHD exhibited lower IC and effortful control, the broader temperamental regulatory domain which encompasses IC of behavioral responses as well as voluntary management of attentional control processes in service of self-regulation (Rothbart, 2007). Further, lower IC (i.e., effortful control) was predictive of both internalizing and externalizing symptoms in both groups of children, though the effect was more pronounced for children with ADHD (DePauw & Mervielde, 2011). Inhibitory control has also been shown to predict subclinical levels of hyperactivity three years later in school-aged children, suggesting IC may be a sensitive early marker of ADHD symptoms (Thorell et al., 2004). Inhibitory control deficits central to ADHD can also differentiate preschoolers with ADHD from those with oppositional defiant disorder or anxiety (Skogan et al., 2015). Further, it is likely that poor IC early in life may signal genetic risk for later ADHD-related problems, as phenotypic covariance between parent-rated IC and ADHD symptoms at 24 months has been largely attributed to common genetic underpinnings (Gagne et al., 2011). Similar to those findings, heritability estimates of IC in middle childhood, a period when many behavioral disorders become apparent, suggest that IC impairment may be characteristic of a particular subgroup of individuals with a familial form of ADHD (Crosbie et al., 2008; Nigg et al., 2004).

ASD.

Because past research involving youth with ASD has mostly focused on global executive deficits or specific impairments in mental flexibility or shifting, the presence of IC differences in ASD is less clear than research on ADHD (Gardiner et al., 2017; Gioia et al., 2002; Hill, 2004; Konstantareas & Stewart, 2006; Smithson et al., 2013; Zantinge et al., 2017). Consistent with the aims of the current paper, previous studies have found that parent-rated IC (i.e., effortful control) differentiated children with and without ASD and predicted ASD symptoms longitudinally (Garon et al., 2016; Konstantareas & Stewart, 2006). Gardiner et al. (2017) also determined that, in a sample of children with ASD, poor IC was related to greater ASD symptom severity. Other studies documenting lower parent-rated IC in children, however, have failed to detect an association between IC and ASD symptoms (Bishop & Norbury, 2005). Sex differences in IC may also differ in the context of ADHD and ASD. Whereas NT males evidence poorer IC ratings and performance relative to female children and adolescents, females with ASD demonstrate poorer response inhibition relative to males with ASD and NT females (Eisenberg et al., 2005; Lemon et al., 2011; Moilanen et al., 2010; Thorell et al., 2004). For instance, female children and adolescents show slower and less accurate response inhibition, an effect moderated by ADHD trait scores, with greater differences in younger females compared to males (Crosbie et al., 2013).

Genetic Model of IC Impairment: Fragile X Syndrome

Fragile X syndrome (FXS) is a single-gene disorder identified by an expansion of the CGG repeat sequence on the long arm of the X chromosome. Though cognitive and behavioral profiles vary across individuals with FXS, the majority of males with FXS have intellectual disability, and the phenotype is also characterized by significant social impairments, internalizing and externalizing behaviors, and elevated rates of ADHD and ASD (Bailey et al., 2008; Baumgardner et al., 1995; Center for Disease Control [CDC], 2021; Hatton et al., 2002; Roberts et al., 2020; Sullivan et al., 2006). Specifically, as many as 61% of preschoolers with FXS meet DSM-5 criteria for ASD, and nearly 80% experience ADHD-related symptoms (Bailey et al., 2008; Roberts et al., 2020). Additionally, ADHD and ASD symptoms commonly co-occur in individuals with FXS, similar to what is seen in the general population (Sinzig et al., 2009; Crawford et al., 2018; Ames & White, 2011; Visser et al., 2016; Smith et al., 2012). The presence of one or both of these co-morbidities is associated with reduced quality of life for children with FXS and their families (Bailey et al., 2008).

Inhibitory control delays are central to the behavioral phenotype of young males with FXS. Further, IC differences documented in young males with FXS cannot be fully explained by general developmental delays (Cornish et al., 2007, 2013; Hooper et al., 2008; Kapalu & Gartstein, 2016; Munir et al., 2000; Scerif et al., 2005, 2007; Sullivan et al., 2007; Tonnsen et al., 2015; Wilding et al., 2002). This is consistent with findings of children with intellectual disability demonstrating poorer verbal and nonverbal IC relative to mental age (MA)- and chronological age (CA)-matched peers (Danielsson et al., 2012; Spaniol & Danielsson, 2021). Males with FXS have exhibited lower performance relative to MA-matched males on various IC-related tasks, including distractor interference, delay of gratification, stop signal, and antisaccade tasks (Cornish et al., 2007; Munir et al., 2000; Scerif et al., 2007; Tonnsen et al., 2015). Still, little is understood about how IC develops very early in male children with FXS. Prior work suggests that IC delays are likely present by toddlerhood in males with FXS, increasing over time (Robinson et al., 2008). Specifically, NT males demonstrate growth in IC (i.e., effortful control) over time, whereas children with FXS evidence flat trajectories (Robinson et al., 2018). Further, for children with FXS, low levels of IC were associated with greater ASD symptoms (Robinson et al., 2018), highlighting the importance of examining IC development as an early indicator of ASD. However, despite clear evidence of IC differences in young males with FXS, to date, no study has specifically examined IC longitudinally in early childhood or investigated IC as a predictor of later behavioral symptoms (e.g., ADHD, ASD) in male children with FXS.

Present study

In sum, it remains unclear how IC develops across early childhood or the cascading impacts of early IC difficulties in males with FXS. The present study aimed to shed light on this question by examining IC abilities longitudinally in males with FXS and NT males between the ages of 16 to 71 months. The primary aim of the study was to characterize early IC trajectories and determine when IC differences become apparent in young males with FXS relative to NT peers. We hypothesized that, compared to their NT peers, male children with FXS would exhibit early emerging IC differences which would become more striking across early childhood. The second aim of the study was to determine whether early levels and/or changes in IC over time were related to ADHD and ASD symptom severity at age 5 in both groups. We hypothesized that IC growth across early development would be associated with both ADHD and ASD symptoms in young males with FXS.

Method

Participants

Sample 1:

To characterize IC trajectories across early development and determine when IC differences become apparent for children with FXS, we identified a sample of 128 eligible male participants, (FXS: n=79, NT: n=49) using extant data from two longitudinal studies of early development involving children with FXS (R01MH090194; R01MH107573; PI: Roberts). The sample included all children who were assessed between 16 and 71 months (1–5 observations per child; 64% of sample completed 1–2 assessments in this age-range), for a total of 284 observations (FXS: n=172; NT: n=112). This age range was chosen because IC emerges around the end of the first year of life and matures steadily, becoming increasingly coherent over this age range (Kochanska et al., 2000).

Demographic characteristics for Sample 1 are shown in Table 1. Maternal education and income were included for participants with available data. For participants with income data available at multiple assessments, the average income was included. A greater proportion of mothers of NT children completed a bachelor’s degree or higher. Income was similarly distributed across FXS and NT groups.

Table 1.

Sample 1 Demographic Characteristics by Percent of Group

	FXS	NT
Ethnicity (%)
n	79	49
Hispanic/Latino	2.5	0
Not Hispanic/Latino	74.7	63.3
Unknown	22.8	36.7
Race (%)
n	79	49
White	73.4	91.8
Black	3.8	6.1
Hispanic/Latino	1.3	0
American Indian/Alaska Native	1.3	0
More than one race	15.2	2.1
Unknown	5.0	0
Maternal Education (%)
n	56	31
Less than high school	7.1	3.2
High school degree	8.9	6.5
Associate’s degree	7.1	9.7
Some college	17.9	6.5
Bachelor’s degree	19.6	25.8
Some graduate work	14.3	6.5
Graduate degree	25	41.9
Income (%)
n	41	29
$5,000–$25,000	7.3	3.4
$25,001–$50,000	26.8	31
$50,001–$75,000	19.5	20.7
$75,001–$100,000	14.6	13.8
$100,001-$150,000	12.2	17.2
$150,000+	19.5	13.8

Sample 2:

To determine whether early IC trajectories predicted ADHD and ASD symptom severity at age 5, we identified a subsample of n=50 children from Sample 1 who had both IC trajectory data across early childhood (26% of children completed 1–2 assessments; 74% of children completed 3–5 assessments) and an outcome assessment that characterized ADHD and ASD symptom severity at 5 years of age (FXS: n=26, NT: n=24; M_age= 60 months; age range= 54–71 months; see Table 5). The age of outcome was selected as 5 years of age because ASD and ADHD symptoms have been shown to emerge and stabilize by this age (Lord, 1995; O’Neill et al., 2014; Ozonoff et al., 2015; Riddle et al., 2013; Van Daalen et al., 2009).

Table 5.

Sample 2 Participant Characterization for IC Trajectories and Age 5 Outcome Data

	FXS (n=26) M (SD)	NT (n=24) M (SD)	d
CA	60.46 (3.74)	60.13 (4.06)	0.09
IC intercept	3.32 (0.40)	3.39 (0.46)	0.16
IC slope	0.011 (0.0079)	0.016 (0.010)	0.59
MSEL ELC	53.00 (6.03)	96.79 (17.71)	3.37
MSEL NVDQ	51.32 (15.64)	96.10 (11.94)	3.20
CBCL-ADHD problems	8.13 (2.26)	3.58 (2.83)	1.77
ADOS-2 CSS	5.88 (2.44)	1.96 (1.46)	1.94

CA= Chronological age in months

IC intercept= Inhibitory control at 24 months

IC slope= Change in inhibitory control over chronological age

MSEL ELC= Mullen Early Learning Composite standard score

MSEL NVDQ= Mullen Nonverbal Developmental Quotient

CBCL-ADHD= Child Behavior Checklist- Attention Deficit/Hyperactivity Disorder problems subscale raw score

ADOS-2 CSS= Autism Diagnostic Observation Schedule, Second Edition Calibrated Symptom Severity score

Recruitment for the larger longitudinal studies was completed via research, parenting, and social media internet sites along with postings in the community and collaborations among research groups. Inclusion criteria for participation were: a) gestational age of at least 37 weeks, b) English as the household primary language, c) no other known medical conditions. Neurotypical children with a family history of ASD or related disorders (e.g., FXS, tuberous sclerosis) or with a diagnosis of ID, ADHD, and ASD were excluded. FXS status was confirmed via genetic report. Because FXS is less common in females, and because females are generally less severely affected in cognitive, behavioral, and social domains, only males were included in the present study (Bailey et al., 2008; Hatton et al., 2006; Klusek et al., 2014; Lee et al., 2016; Rinehart et al., 2011).

Measures

Inhibitory control (IC).

IC was assessed via the inhibitory control subscale from the Rothbart Scales of Temperament. Parents completed the Early Childhood Behavior Questionnaire (ECBQ; Putnam et al., 2006) for children ages 16 to 36 months and the Children’s Behavior Questionnaire (CBQ; Rothbart et al., 2001) for children ages 36 months and older. The creation of these measures was informed by theory of temperamental continuity, and they have been previously used together in longitudinal examinations of temperament in NT children (Joyce et al., 2016; Posner & Rothbart, 2000; Spinrad et al., 2012). A meta-analysis examining the developmental validity of IC measures supported the utility of the IC subscale of the CBQ in examining individual differences in IC (Petersen et al., 2016). Further, a confirmatory factor analytic study of the CBQ in males with FXS retained a three-factor model of temperament, similar to that demonstrated by Rothbart and colleagues (2001), demonstrating its valid use with this population (Roberts et al., 2014). IC raw scores range from 1–7, with higher scores indicating greater IC ability.

ADHD Symptom Severity.

ADHD symptom severity was measured at the outcome assessment using the Child Behavior Checklist for Ages 1½−5 (CBCL; Achenbach & Rescorla, 2000), a parent rating scale of emotional and behavioral functioning, which has been used previously with children with FXS (Grefer et al., 2016; Hatton et al., 2002; Sullivan et al., 2006). As suggested by the test publishers, raw scores on the DSM-oriented Attention Deficit/Hyperactivity (ADHD) Problems subscale range from 0 to 12 and were used as the outcome variable, with higher scores reflecting more severe ADHD symptoms. This subscale has demonstrated high inter-rater and test-retest reliability (r=.96; r=.93; Nakamura et al., 2009).

ASD Symptom Severity.

ASD symptom severity was assessed at the outcome assessment via the Autism Diagnostic Observation Schedule, Second Edition (ADOS-2; Lord et al., 2012), a play-based, semi-structured socio-behavioral assessment and gold-standard tool for the diagnosis of ASD which has been extensively used to measure ASD symptoms in individuals with FXS (Abbeduto et al., 2019; Klusek et al., 2014; Roberts et al., 2018; Wall et al., 2019). Calibrated severity scores (CSS) were included in analyses as a continuous measure of ASD symptom severity. CSS scores range from 1–10, with higher scores indicating more severe ASD symptoms.

Developmental Level.

Developmental level was measured via the Mullen Scales of Early Learning (MSEL; Mullen, 1995), a standardized assessment of developmental abilities in early childhood. The MSEL Early Learning Composite (ELC) is comprised of Fine Motor, Visual Reception, Receptive Language, and Expressive Language domains and is utilized as an index of general developmental level in the current study. The inclusion of ELC in our statistical models was not supported; however, it is included in the present study for descriptive purposes. Additionally, a nonverbal developmental quotient (NVDQ) was computed as the (average of visual reception and fine motor age equivalents)/chronological age*100.

Procedures

Procedures were approved by the Institutional Review Boards at the University of South Carolina and the University of North Carolina at Chapel Hill. Prior to study enrollment parents provided written informed consent. As part of the larger longitudinal studies, children were assessed at 18, 24, 36, 48, 60, and 72 months using a standard research protocol consisting of various developmental and clinical measures. Assessments took place either in the child’s home or in a research lab setting. Parent- report measures assessing IC and ADHD symptoms were mailed home and completed prior to each assessment. The ADOS-2 was administered by research reliable personnel. ADOS-2 module was determined according to participant age and expressive language abilities. As part of the study protocol, a clinical best estimate (CBE) diagnosis (i.e. ASD, non-ASD developmental delay, no clinical features), adapted from standard procedures (Lord et al., 2012; Lord et al., 2006), was determined based on a review of data by a multidisciplinary team including a licensed psychologist who is a research-reliable trainer for the ADOS-2. Although a small subset of NT males exhibited symptoms of ASD or ADHD as is expected in a community sample, we confirmed that none met diagnostic criteria for ASD or ADHD according to this CBE process (see procedures above). However, NT children with higher or clinically-significant levels of behavioral symptoms according to the CBCL or ADOS-2 were retained in the present sample and regarded as reflecting the true spectrum of typical functioning.

Statistical Analysis Plan

Prior to statistically testing the primary aims of the study, we calculated descriptive analyses to characterize the groups. For Sample 1, groups were matched on CA using a pseudo-randomized matching process based on the grand mean of all observations. This process involved a) assigning a random number to each observation; b) splitting the dataset into quartiles; c) sorting by random number and number of observations available for each participant; and d) systematically trimming to remove cases, resulting in groups with minimal differences in (Cohen’s d= 0.20, p=.112) and similarly distributed CA. Participants from Sample 1 were included in Sample 2 if they had available outcome data (i.e., ADHD symptoms, ASD symptoms) at age 5. Additionally, Sample 2 NT and FXS groups were matched on chronological age (Cohen’s d= 0.08, p=.762). In addition, all data were examined for linearity and normality using data visualization, Q-Q plots, boxplots, and Shapiro-Wilk test of normality (p>.05). Levene’s test was used to assess the assumption of homogeneity of variance. Violations to the assumption of normality as indicated by significant results on the Shapiro-Wilk test was found for ADHD symptoms in the FXS group and ASD symptoms in the NT group. As expected, ASD symptoms violated the assumption of homogeneity of variance across groups. Transformation of skewed variables did not improve the model; therefore untransformed variables were retained. No assumptions were found to be violated for the remaining variables.

Demographic frequencies and descriptive statistics for Sample 1 are included in Tables 1 and 2. Differences between groups on all primary variables of interest were evaluated descriptively via effect sizes (i.e., Cohen’s d) rather than t-tests in order to reduce the likelihood of Type I error through multiple comparisons (Tables 2, 4, and 5). Figure 1 illustrates the distribution of ASD and ADHD symptom levels across groups. Examination of the ECBQ and CBQ IC subscales suggests items do not require functional language; additionally the correlation between IC and the MSEL Expressive Language age equivalent was minimal (r(144)=.17, p=.046), therefore verbal ability was not deemed a necessary covariate. Bivariate correlations were computed to determine whether to control for nonverbal cognitive ability (i.e., MSEL NVDQ). There was a moderate, significant correlation between IC and NVDQ in the FXS group, (r(170)=.37, p<.001), and a small, significant correlation the NT group, (r(110)=.22, p=.018). Consequently, NVDQ was included as a covariate in the initial linear mixed model. Because the inclusion of NDVQ as a covariate in the initial longitudinal models accounted for effects of NVDQ on IC slopes and intercepts extracted for use in subsequent models, NDVQ was not including in any further models (i.e., Sample 2).

Table 2.

Sample 1 Descriptive Statistics by Age

	FXS M (SD)	NT M (SD)	d
All ages*
n	172 observations	112 observations
CA	45.29 (13.95)	42.48 (14.85)
IC	3.11 (0.82)	4.32 (0.97)	1.37
18 months
n	7	13
CA	18.86 (1.21)	18.15 (0.55)
ECBQ IC score	2.89 (0.56)	3.60 (1.02)	0.80
24 months
n	22	12
CA	25.18 (1.22)	24.83 (1.11)
ECBQ IC score	3.68 (0.92)	3.76 (0.90)	0.09
36 months
n	44	28
CA	37.09 (2.01)	36.89 (1.62)
CBQ IC score	2.86 (0.78)	4.34 (0.91)	1.78
48 months
n	55	32
CA	49.16 (3.37)	47.41 (3.30)
CBQ IC score	3.07 (0.81)	4.46 (0.80)	1.74
60 months
n	44	27
CA	62.91 (5.38)	62.00 (5.14)
CBQ IC score	3.15 (0.75)	4.71 (0.98)	1.85

^*=Pooled mean and SD across repeated observations

CA= Chronological age in months

ECBQ= Early Childhood Behavior Questionnaire

IC= Inhibitory control

CBQ= Children’s Behavior Questionnaire

Table 4.

Sample 1 Extracted Subject-Specific IC Trajectory Parameters by Group

	FXS (n=79) M (SD)	NT (n=49) M (SD)	d
IC intercept	3.32 (0.35)	3.38 (0.45)	0.15
IC slope	0.011 (0.0063)	0.015 (0.0078)	0.47

IC intercept= Inhibitory control at 24 months

IC slope= Change in inhibitory control over chronological age

Figure 1.

Variance in Psychiatric Symptom Severity at Age 5 by Group

Note. Scores above the red lines fall within the borderline clinical or clinical range for the CBCL, and autism spectrum disorder or autism range for the ADOS-2.

To characterize IC growth across early development and determine the age at which IC differences become statistically significant in young males with FXS, we employed a linear mixed model using the lme4 package in R (Bates et al., 2015; RStudio Team, 2016) examining IC across age (24–71 months). A linear model was chosen given the majority of participants (64%) completed 1–2 assessments and data visualization indicated a linear relationship between IC and chronological age, which is consistent with recent studies examining IC development across this age range (Hooper et al., 2018; Hughes & Ensor, 2011; MacDonald et al., 2014; Tonnsen et al., 2015). This type of modeling allows for within- and between-subject variance over time and is flexible to variable timing and frequency of assessments, as is the case in our sample (Singer, 1998). Similar statistical approaches have been used with other small samples (Baranek et al., 2008; Hazlett et al., 2012; Neuhaus et al., 2016). Group, age, NVDQ, and a group*age interaction were included as predictors, with age specified as a level-1 predictor nested within participant and centered at the earliest timepoint with a sizeable number of observations (24 months) and group as a level-2 predictor. Random effects were estimated for the intercept and slope to allow for individual variability in initial levels of IC and change in IC over time.

To examine whether IC at 24 months (intercept) or developmental trajectories of IC (slope) across early childhood are associated with ADHD and/or ASD symptom severity at age 5, individual regression lines were fit to the longitudinal IC data for each participant and subject-specific IC slope and intercept were extracted (Table 4). We employed partial correlation analyses to determine whether IC intercept or IC slope was a more salient predictor of specific neurodevelopmental outcomes (i.e., ADHD, ASD symptoms). We chose a partial correlation approach because it estimates the unique variance accounted for by each predictor in each outcome, while removing the effects associated with the other parameter (i.e. IC slope, IC intercept).

Results

Descriptive Analyses

Compared to the NT group, the FXS group exhibited lower IC at each time point (Cohen’s d range= 0.09–1.85; see Figure 2). Based on averaged parameters extracted from individual regression lines, the FXS group demonstrated lower IC slope and intercept in both Samples 1 and 2 (Tables 4 and 5).

Figure 2.

Comparison of IC Trajectories by Group

For Sample 2, Cohen’s d values indicated a large mean difference between groups such that, as anticipated, the FXS group demonstrated lower developmental ability (Cohen’s d= 3.37) and nonverbal developmental ability (Cohen’s d= 3.20), and higher levels of ADHD (Cohen’s d= 1.77) and ASD symptoms (Cohen’s d= 1.94) at age 5 compared to NT males (Table 5).

IC Maturation Across Early Development

A linear mixed model tested the effect of group on IC across age (see Table 3). Results revealed a significant main effect of NVDQ, (b=0.014, t(236.4)=3.56, p<.001), indicating a significant association between NVDQ and IC for both FXS and NT males at 24-months old, but no main effect of group, suggesting that groups do not differ in levels of IC at 24 months when controlling for NVDQ, (b=0.25, t(120.7)=1.05, p=.296). A significant age-by-group interaction, (b=0.018, t(75.18)=2.60, p=.011), was also identified, suggesting a difference in the slope of IC maturation between males with FXS and NT males. Specifically, NT males increased in IC, whereas males with FXS, on average, demonstrated decreasing levels of IC over time (see Figure 2), suggesting an increased magnitude of group differences in IC across development. Given groups did not significantly differ at 24 months but the age-by-group interaction was significant, we probed the interaction to determine the age at which divergence in IC became statistically significant. To probe the interaction, we recentered age at 36-months (the next timepoint) and retested the model. These model results indicated a significant main effect of group, (b=0.46, t(178.9)=2.12, p=.035), suggesting group differences in IC emerge by 36 months.

Table 3.

Sample 1 Linear Mixed Model Results of IC Trajectories in FXS and NT Centered at 24 Months

	Estimate	SE	df	t	p
Intercept	3.29	0.13	98.85	25.72	<.001
Age	0.0037	0.0045	97.31	0.82	.413
Group	0.25	0.24	120.7	1.05	.296
NVDQ	0.014	0.0041	236.4	3.56	<.001
Age × Group	0.018	0.0069	75.18	2.60	.011

Note. FXS set as reference group in model.

Associations Between IC and ADHD & ASD Symptoms at Age 5

FXS Group.

Partial correlation analyses were tested to determine the relationship between each IC parameter (i.e., slope, intercept) and ADHD symptoms, controlling for the other parameter. Controlling for IC intercept, there was a small significant relationship between IC slope and ADHD symptoms, (r_partial = −.47, p=.022). In addition, when controlling for IC slope, the partial correlation between IC intercept and ADHD symptoms, (r_partial = −.58, p=.004), was significant, suggesting the association between IC intercept and ADHD is not better explained by the association between either of these variables with IC slope (Figure 4). Based on the magnitude of these associations, these results suggest that greater ADHD symptoms in males with FXS have a slightly stronger association to low levels of IC at 24 months (i.e., IC intercept) relative to the association between decreased slopes in IC over time and ADHD symptoms.

Figure 4.

Comparison of IC Intercept Predicting ADHD Symptoms by Group

Partial correlation analyses revealed a significant correlation between IC slope and ASD symptoms, (r_partial = −.50, p=.010), suggesting the association between IC slope and ASD symptoms is not better explained by the association between either of these variables with IC intercept (Figure 5). However, when controlling for IC slope, the partial correlation between IC intercept and ASD symptoms was non-significant, (r _partial = −.08, p=.703; see Figure 6). These findings indicate that a decreased slope in IC across early development is associated with greater ASD symptoms at age 5 in young males with FXS, whereas low levels of IC at 24 months (i.e., IC intercept) is not significantly related to ASD symptom severity in this group.

Figure 5.

Comparison of IC Slope Predicting ASD Symptoms by Group

Figure 6.

Comparison of IC Intercept Predicting ASD Symptoms by Group

Neurotypical Group.

Partial correlation analyses were run to determine the relationship between each IC parameter (i.e., slope, intercept) and ADHD symptoms, controlling for the other parameter. Results of the partial correlations identified a strong and significant correlation between IC slope and ADHD symptoms (Figure 3), controlling for IC intercept, (r_partial = −.66, p<.001), suggesting the association between IC slope and ADHD symptoms is not better explained by the association between either of these variables with IC intercept. The association between IC intercept and ADHD symptoms, controlling for IC slope, was also significantly and strongly correlated, (r_partial = −.69, p<.001), suggesting the association between IC intercept and ADHD symptoms is not better explained by the association between either of these variables with IC slope (Figure 4). Similar to findings for the FXS group, these results suggest that higher levels of IC at 24 months (i.e., IC intercept) and an increased slope in IC over time are both associated with lower ADHD symptoms in NT male children. For NT males, when accounting for IC intercept, the partial correlation between IC slope and ASD symptoms was non-significant, (r _partial = −.21, p=.323; see Figure 5). Similarly, when accounting for IC slope, the partial correlation between IC intercept and ASD symptoms was non-significant in the NT group, (r _partial = .30, p=.161; see Figure 6). These findings suggest no association between early levels or growth in IC across early childhood with ASD symptoms in NT male children.

Figure 3.

Comparison of IC Slope Predicting ADHD Symptoms by Group

Discussion

The present study aimed to determine whether and when young males with FXS differ from their NT peers in parent ratings of IC. To our knowledge, this is the first study to specifically map within-individual trajectories of IC across the toddler and preschool years in male children with FXS relative to NT male children, as well as to examine the associations between these trajectories and different behavioral symptoms in these groups. Overall, findings indicate that group differences in IC abilities among young males with FXS and their NT peers emerge between 24 and 36 months of age and these differences widen across early childhood. Additionally, whereas low early levels and a decreased slope in IC across early childhood are associated with greater ADHD symptoms at age 5 in neurotypical development, low levels of IC at 24 months are related to greater ADHD symptoms and a decreased slope in IC is associated with greater ASD symptoms at age 5 in young males with FXS.

In the present study, controlling for nonverbal developmental ability, males with FXS demonstrated lower IC than NT males by 36 months of age. This confirms that IC is in fact delayed in boys with FXS, representing a true neurocognitive delay. These findings are consistent with previous reports of impaired IC performance in school-aged males with FXS (Cornish et al., 2007; Hooper et al., 2008; Munir et al., 2000; Scerif et al., 2007; Sullivan et al., 2007; Wilding et al., 2002), but demonstrate the emergence of IC delays earlier in development. In addition to evidence of IC divergence from typical development by 36 months of age, many young males with FXS also demonstrated decreases in IC skills across age, with the group mean trajectory showing decreases in IC across early development (Figure 1). This is consistent with prior work showing stagnant growth in regulatory abilities in males with FXS (Robinson et al., 2018). These findings also extend previous work demonstrating slow yet positive EF growth with increasing mental age across childhood and adolescence in males with FXS (Cornish et al., 2013; Hooper et al., 2018; Tonnsen et al., 2015). In summary, results indicate that the IC delays in males with FXS are evident as early as the third year of life with demonstrated divergence from typical development and decreased slopes over time.

Early IC difficulties and a lack of improvement in IC across early childhood likely set the stage for several social and behavioral impairments central to the FXS phenotype, such as low social competence, difficulties with self-regulation, and poor adjustment outcomes such as increased internalizing and externalizing symptoms. Our results indicated that lower IC levels at 24 months were associated with greater ADHD symptoms at age 5 in young males with FXS and NT males. Further, 24-month IC levels were associated with age 5 ADHD symptoms in young males with FXS regardless of whether their IC improved, stagnated, or declined over time. These findings are seemingly at odds with evidence of increasing expression of ADHD symptoms with age in individuals with FXS (Grefer et al., 2016). A strong relationship between early levels of IC and ADHD symptoms at age 5 was also seen in the NT group, which is what we would expect based on the existing literature. That these associations were not unique to the FXS group and were evident in the context of sub-clinical levels of ADHD symptoms for the NT group suggests that IC is a central construct to ADHD across groups. Collectively our results suggest that early delays in IC at 24 months may serve as a prognostic marker for later ADHD-related symptomatology in males with FXS, replicating evidence of longitudinal predictive validity and specificity of IC to ADHD for children with both low and high familial risk for ASD (Brocki et al., 2007; Campbell & Von Stauffenberg, 2009; Shephard et al., 2019). Ultimately, our findings suggest a broader ADHD subtype in young males with FXS that is characterized by neurocognitive delays, rather than a model of heterogeneity, in which only some individuals with ADHD demonstrate IC impairments (Sonuga-Barke 2002; Nigg et al., 2004). For the NT group, higher IC levels at 24 months and increased slopes in IC across childhood were both related to lower ADHD symptom severity at age 5. Interestingly, despite lower levels of ADHD symptoms on average in the NT group compared to the FXS group, the magnitude of association suggests both IC levels (i.e., r= −.69) and growth (i.e., r= −.66) are moderately associated with ADHD symptoms in the NT group.

Regarding ASD symptoms, findings suggest that IC trajectories, but not initial IC levels, were related to ASD symptoms in male children with FXS; further, this association was independent of initial IC levels and NVDQ. These results reflect a more nuanced association between IC and ASD than has been previously described (Gardiner et al., 2017; Konstantareas & Stewart, 2006). Previous studies reporting no significant associations between age-related improvements in regulatory abilities and ASD symptoms for children with FXS (Robinson et al., 2018) may be explained in part by limited stability in ASD symptoms at the age of assessment. In contrast, the present study included behavioral symptom levels at a clinically stable outcome age. Contrasting findings in the neurotypical development literature (Gardiner et al., 2017), our results indicated decreasing linear trends in IC across age in the FXS sample; however, these findings are consistent with evidence of more apparent EF deficits with increasing age for children with ASD (Hill, 2004). Although a certain threshold of baseline IC is essential to daily functioning, our results suggest that decreasing trajectories of IC in young males with FXS, rather than low levels of IC at 24 months, may contribute to elevated rates of ASD symptoms in this population. These findings complement existing research indicating that ASD symptoms manifest and intensify in severity with increasing age for children with FXS (Lee et al., 2016), with reduced regulatory ability and increased age-related social-communication demands interacting and culminating in greater externalization of symptoms. Further, early IC impairment and increased impairment in IC across early childhood likely have implications for the development of other complex, higher-order functions and social behaviors. In the NT group, ASD symptoms were unrelated to IC levels at 24 months or growth in IC with age. Low levels and limited variability of ADOS-2 CSS scores in the NT group may in part explain these null findings. Alternatively, given children with higher ADOS-2 CSS scores are more likely to be diagnosed with ASD, findings may suggest that IC less associated with ASD symptoms at subclinical levels. The present study illustrates the early emergence of IC impairment prior to the average age of diagnosis of ASD or ADHD, potentially implicating the primacy of IC delays or mechanistic involvement of this neurocognitive ability in the development of these neurodevelopmental disorders in young males with FXS (Daniels & Mandell, 2014; Visser et al., 2014); however, confirmation of IC as a mechanistic underpinning would require future work to control for early levels of ASD or ADHD symptoms. It is evident that delayed IC, and particularly blunted or decreasing trajectories of IC across early childhood, has cascading effects for individuals with FXS including heightened vulnerability for ADHD and ASD, disorders that impair daily functioning and reduce quality of life for these individuals and their families (Hatton et al., 2002). It is important to note that the present study examined ADHD and ASD symptoms continuously and not formal diagnoses. Further work is needed to clarify whether IC is related to the diagnosis of ADHD or ASD for individuals with FXS or may contribute to the differentiation of these neurodevelopmental disorders from the FXS phenotype. Regardless, the knowledge gleaned from the present study may identify IC as a developmentally sensitive target for early and individualized intervention of these behavioral symptoms, which we know to be important for ensuring optimal developmental outcomes for these children. Further, the lack of specificity of IC in FXS, as indicated by significant associations between IC and ADHD symptoms in NT children, suggests treatments already developed to address ADHD may have similar effectiveness for children with FXS. Studies implementing motoric stopping and proactive monitoring of contextual cues trainings indicated both strategies successfully improved response inhibition (Chevalier et al., 2014; Traut et al., 2021). Additionally, Traut et al. (2021) determined that children with low proactive control, defined as maintenance of goal-relevant information, benefitted from monitoring training; this type of training may also be appropriate for children with low cognitive ability or high levels of inattention. A review of executive function (EF) interventions for children determined that children with the lowest initial EF levels evidence the greatest gains following these interventions (Diamond & Lee, 2011), findings which are promising for children with FXS. Interventions that alter deviant IC trajectories may reduce the severity or prevent the development of later behavioral symptoms in young males with FXS and ultimately reduce emotional stress and economic burden associated with caring for children with developmental disabilities.

This study is the first longitudinal study of early IC trajectories in relation to neurodevelopmental outcomes in a sample of children with FXS and represents a large sample size relative to most research involving rare genetic disorders, though it is not without limitations. Given the majority of participants completed fewer than 3 assessments, it was not possible to model nonlinear trends in the present study, which may have indicated differential growth rate (i.e., acceleration or deceleration) in IC at different points in the early childhood period. Future longitudinal studies should examine potential nonlinear trends in IC development, which may have important implications for timing of interventions prior to steep declines in IC.

The present study leveraged parent-ratings of child IC and demonstrated that IC difficulties are not only early emerging in males with FXS but are also salient to parents at an early age despite the complex behavioral phenotype evident in individuals with FXS. This suggests that parent-report measures of IC are sensitive to early group differences. Additionally, though parent ratings have the benefit of high ecological validity from having observed a child’s behaviors across a variety of settings, these ratings may be biased by knowledge of their child’s diagnosis or differential expectations for their child’s behavior at certain ages. This may be influenced by level of parental education, access to medical and mental health resources, and having older children with developmental disabilities. One limitation of the present study is that the FXS and NT samples are not matched on demographic information (i.e., maternal education, income), and therefore it is possible that these characteristics may impact parent ratings. It is possible that shared method invariance partially, but likely not completely, accounts for the association between IC and ADHD (Nigg, Goldsmith, et al., 2004); however, significant relationships between IC and ASD symptoms, as measured by a direct observational measure, support the validity of our findings and reduce the likelihood of shared method invariance driving results. Characterization of the centrality of early IC impairment to FXS may prime parents of children with FXS to monitor this ability and to pursue early intervention services specific to regulatory vulnerabilities, which may mitigate the development of maladaptive behaviors for children with FXS.

Currently, there are few validated IC tasks for use with young children with intellectual disability. The development of behavioral performance measures of IC for use with young children that are sensitive to “extreme” ability levels could improve understanding of developing brain-behavior relationships and facilitate the examination of heterogeneous mechanistic pathways involved in disorders characterized by executive deficits (Blair et al., 2005). Accordingly, next steps of this work should entail inclusion of a matched comparison group of males with another neurodevelopmental disorder (e.g. Down syndrome, idiopathic autism). One weakness of the present study is the inclusion of only male children. Despite evidence of poorer IC ratings and performance by NT male compared to female children and adolescents and atypical neural activation patterns underlying IC in females with FXS, few studies to date have examined IC in females with FXS or sex differences in these abilities in individuals with FXS (Eisenberg et al., 2005; Garon et al., 2016; Menon et al., 2004; Moilanen et al., 2010; Thorell et al., 2004). Though the present study did not examine sex effects given the lack of an adequate sample of young females with FXS, research on IC as an early behavioral marker in females with FXS should also be explored. Finally, examination of IC functioning at multiple levels of analysis (i.e. genetic, neural, physiological, cognitive, behavioral) will provide a more comprehensive understanding of this neurocognitive ability in individuals with FXS.

In conclusion, the present study determined that compared to their NT peers, young males with FXS exhibit IC delays before 36 months and decreased IC trajectories through age 5. We also determined that IC levels at 24 months are associated with later ADHD symptoms in young males with FXS. Alternatively, a decreased slope in IC over time was associated with greater risk for ASD symptoms in male children with FXS. Differentiating the mechanisms leading to these neurodevelopmental outcomes may enable implementation of more specific interventions that are effective in promoting optimal outcomes for individuals. Our findings help to refine the early developmental phenotype of FXS and highlight the utility of monitoring change in developmental abilities over time. Finally, the present study points to IC as a potential target for early detection and intervention of ADHD and ASD symptoms in male children with FXS.

Table 6.

Sample 2 FXS Group Bivariate (and Partial) Correlations

	1	2	3	4
1. IC intercept	-
2. IC slope	−.09	-
3. CBCL-ADHD	(−.58*)	(−.47†)	-
4. ADOS-2 CSS	(−.08)	(−.50†)	.26	-

IC intercept= Inhibitory control at 24 months

IC slope= Change in inhibitory control over chronological age

CBCL-ADHD= Child Behavior Checklist- Attention Deficit/Hyperactivity Disorder problems subscale raw score

ADOS-2 CSS= Autism Diagnostic Observation Schedule, Second Edition Calibrated Symptom Severity score;

Partial correlations, controlling for the other IC parameter, are shown in parentheses;

^†p<.05,

^*p<.01,

^**p<.001

Table 7.

Sample 2 NT Group Bivariate (and Partial) Correlations

	1	2	3	4
1. IC intercept	-
2. IC slope	.11	-
3. CBCL-ADHD	(−.69**)	(−.66**)	-
4. ADOS-2 CSS	(.30)	(−.21)	−.03	-

IC intercept= Inhibitory control at 24 months

IC slope= Change in inhibitory control over chronological age

CBCL-ADHD= Child Behavior Checklist- Attention Deficit/Hyperactivity Disorder problems subscale raw score

ADOS-2 CSS= Autism Diagnostic Observation Schedule, Second Edition Calibrated Symptom Severity score;

Partial correlations, controlling for the other IC parameter, are shown in parentheses;

^†p<.05,

^*p<.01,

^**p<.001