Participants

This cross-sectional study included a convenience sample of 243 (48% female) children and young people with 1 of 8 genetically confirmed monogenic NDDs (pertaining to variants in CDK13, DYRK1A, FOXP2, KAT6A, KANSL1, SETBP1, BRPF1, and DDX3X) aged 1 to 25 years (mean = 8 years 10 months, SD = 5 years 8 months). Families were recruited nationally and internationally through the National Health and Medical Research Council (Australia) Centre of Research Excellence in Speech and Language, as part of broader cohort studies of these conditions from January 2019 to December 2023. These broader ‘reverse phenotyping’ cohort studies aimed to provide comprehensive phenotyping of the speech and language profiles of several genetic conditions.^{12, 15-21} Individuals were included in these previous studies if they had a confirmed molecular diagnosis and excluded if they had any other confirmed genetic variant or syndrome that was likely to affect the clinical phenotype.

Participants were included in the current study based on having a completed Vineland Adaptive Behavior Scales, Third Edition (Vineland-3)²² and aged between 1 year and 25 years. Ethics approval was obtained from the Royal Children's Hospital Human Research Ethics Committee (no. HREC37353). Informed written consent was provided by caregivers and legal guardians.

Data collection

Parents completed an online health and medical questionnaire to collect demographic information and determine the presence of co-occurring conditions, such as intellectual disability, autism spectrum disorder (hereafter ‘autism’), and attention-deficit/hyperactivity disorder (ADHD). Response options for autism and ADHD were ‘yes’ or ‘no’, while for intellectual disability the response options were, ‘yes’, ‘no’, or ‘not applicable’ (too young or not tested). When indicated to be present, parents and caregivers uploaded copies of original clinical assessment reports, if applicable, to confirm the presence or absence of the condition (intellectual disability, autism, and ADHD). In some instances, no formal assessment was completed but parents indicated the presence or absence of the condition. For other individuals, no formal assessment or parent report of a diagnosis was provided. For these individuals, it was assumed that there was no indication or parental concern of autism or ADHD that warranted assessment; thus, the condition was classified as ‘absent’. To enable classification of intellectual disability in these individuals, review of the health and medical history form, including age at attaining developmental milestones, presence or absence of developmental delay, and letters from medical or allied health professionals, was undertaken by the first author (EKB) to provide a ‘clinician-inferred’ presence or absence of intellectual disability.

Parents completed the Vineland-3 comprehensive parent form²² to assess communication, daily living, socialization, and motor skills, and their respective subdomains. The Vineland-3 was completed in the participants' preferred language (English, Spanish, Dutch, or German) through an online research data collection platform (REDCap, Vanderbilt University, Nashville, TN, USA)^{23, 24} hosted at the Murdoch Children's Research Institute, with permission to reproduce the Vineland-3 on REDCap from Pearson Clinical Assessments Australia. Scaled scores for the Adaptive Behaviour Composite (ABC) and the Vineland-3 domains were calculated, as were the v-scaled scores for the subdomains. Scaled scores ranged from 20 to 140, with a mean of 100 and SD of 15, while v-scaled scores ranged from 1 to 24, with a mean of 15 and SD of 3. Qualitative descriptors are also provided for ranges of scaled scores (20–70 = low; 71–85 = moderately low; 86–114 = adequate; 115–129 = moderately high; 130–140 = high) and v-scaled scores (1–9 = low; 10–12 = moderately low; 13–17 = adequate; 18–20 = moderately high; 21–24 = high).

Statistical analysis

Demographic (age and sex) and clinical characteristics (intellectual disability, autism, and ADHD) were summarized between diagnostic groups. Frequencies and percentages were used for categorical variables and means and SDs for child age. Linear regression models compared mean adaptive behaviour skills between diagnostic groups, adjusting for the presence of intellectual disability. Model-adjusted means and 95% confidence intervals (CI) were plotted; post hoc Bonferroni-corrected comparisons were made between each group. Overall, data missingness was low for adaptive behaviour skills outcomes (4.9%)²⁵ and results presented using all available data.

To examine potential underlying adaptive behaviour skills profiles in the group, latent profile analysis (LPA) was conducted on the total sample. LPA is a mixture model used to determine latent groups of a sample from observed variable patterns.²⁶ Observed v-scaled scores for the subdomains of communication, daily living, and social skills were standardized, and the LPA of these with two to seven profiles were compared for optimal fit. Model fit was assessed using the Akaike and the Bayesian information criteria, then the Vuong–Lo–Mendell–Rubin adjusted likelihood ratio test and entropy. A better model fit was indicated by lower Akaike and Bayesian information criteria values, with emphasis placed on a change (Δ) in Bayesian information criteria greater than 10, that is, ‘strong evidence’ for improvement,²⁷ a significant Vuong–Lo–Mendell–Rubin p-value, and entropy greater than 0.8 (with values closer to 1 being optimal).²⁶ Demographic and clinical characteristics, including autism, ADHD, and intellectual disability were again compared between the derived profiles. Statistical significance was set at p < 0.05 for all analyses; all analyses were carried out using Stata v18.0 (StatCorp, College Station, TX, USA).²⁸

Co-occurring conditions

Of the 243 participants, 73 (30%) had a confirmed intellectual disability based on documented clinical assessment. A further 61 (25%) individuals were reported by their parent or caregiver to have an intellectual disability, and 48 (20%) had a clinician-inferred intellectual disability. Only 21 (9%) participants were confirmed to have no intellectual disability based on documented clinical assessment, while a further 21 (9%) were reported by their parents to have no intellectual disability, and 8% were clinician-inferred as not having an intellectual disability. Overall, 182 (75%) of the children and young adults in this study were deemed to have an intellectual disability.

Regarding the presence of other co-occurring NDDs, 32 (13%) participants had a confirmed diagnosis of autism, with a further 17 (7%) reported to have a diagnosis of autism by their parent or caregiver. Similar rates of ADHD were endorsed, with 24 (10%) participants confirmed to have a diagnosis of ADHD and a further 14 (6%) with a diagnosis of ADHD according to parent report.

For most of the genetic conditions, the presence of intellectual disability was confirmed or parent-reported, except for individuals with the BRPF1 or FOXP2 variants (Table 1). Most children were not indicated as having autism or ADHD, except for those with a DYRK1A variant, where almost 50% of children either had a confirmed diagnosis or parent report of autism. Information relating to the age, biological sex, intellectual ability, and presence of other co-occurring conditions (autism and ADHD) are provided in Table 1.

Syndrome-specific profiles and group comparisons of adaptive behaviour skills

Examining the adaptive behaviour profiles of each individual syndrome, skills were homogeneous across the domains assessed (Figure 1). For most syndromes, socialization (KANSL1, CDK13, DDX3X, KAT6A) or motor skills (FOXP2, SETBP1, DYRK1A) were the strongest skillset in the profile, while communication skills were the weakest, except for the KANSL1 and CDK13 groups where daily living skills were the weakest. Means and SDs for the ABC, domain scores, and subdomain scores for each genetic syndrome are provided in Table S1.

Few group differences existed when comparing the domain scores of the Vineland-3, adjusting for intellectual disability. The BRPF1 group had significantly (post hoc pairwise Bonferroni comparisons) higher ABC and daily living scores compared to the CDK13 (ABC: mean difference [M_diff] = 15.5, 95% CI = 0.3–30.6; daily living: M_diff = 18.4, 95% CI = 1.5–35.3), DDX3X (ABC: M_diff = 19.5, 95% CI = 2.1–37.0; daily living: M_diff = 20.4, 95% CI = 1.0–39.8), DYRK1A (ABC: M_diff = 21.4, 95% CI = 5.1–37.7; daily living: M_diff = 23.9, 95% CI = 5.8–42.0), and KAT6A (ABC: M_diff = 21.8, 95% CI = 5.5–38.2; daily living: M_diff = 20.6, 95% CI = 2.9–38.3) groups. The BRPF1 group also had significantly higher communication scores compared to the DDX3X (M_diff = 25.5, 95% CI = 3.6–47.4) and KAT6A (M_diff = 23.3, 95% CI = 3.2–43.3) groups (Figure 1).

The KANSL1 group had higher ABC, communication, and socialization scores than the DYRK1A (ABC: M_diff = 13.0, 95% CI = 3.5–22.5; communication: M_diff = 15.3, 95% CI = 3.4–27.3; socialization: M_diff = 18.1, 95% CI = 6.3–29.9) and KAT6A (ABC: M_diff = 13.4, 95% CI = 3.9–22.9; communication: M_diff = 18.5, 95% CI = 7.1–30.0; socialization: M_diff = 17.5, 95% CI = 6.2–28.7) groups. Additionally, the KANSL1 group had higher communication scores compared to the DDX3X (M_diff = 20.8, 95% CI = 6.6–34.9) group, and higher daily living scores compared to the DYRK1A (M_diff = 11.3, 95% CI = 0.7–21.9) group. The SETBP1 group had higher daily living skills compared to the DYRK1A (M_diff = 14.8, 95% CI = 2.6–27.0) group. No other statistically significant between-group differences were detected. The pattern of group differences was similar for the subdomain scores (Figure S1).

Transdiagnostic approach to adaptive behaviour profiles

An LPA compared two to five latent profile models and showed robust evidence for a five-profile model as the best model fit (Table S2). The profiles were homogeneous, with similar delays across the subdomain scores (Table 2). Consequently, profiles were labelled ‘extremely impaired’ (profile 1), ‘very impaired’ (profile 2), ‘moderately impaired’ (profile 3), ‘mildly impaired’ (profile 4), and ‘within normal limits’ (profile 5). Examining the genetic conditions that fell within each profile revealed within-syndrome heterogeneity, with most conditions having individuals falling within each profile (Figure 2), the exceptions being BRPF1 (no individuals in profiles 1 and 2), SETBP1 (no individuals in profile 1), and DDX3X (no individuals in profile 5).

The contribution of autism, ADHD, and intellectual disability to the specific LPA profiles was examined. A higher proportion of individuals with co-occurring autism were in the extremely and moderately impaired profiles compared to the mildly delayed and within normal limits profiles. No specific pattern for ADHD was observed across the profiles. Regarding intellectual disability, a higher proportion of individuals with intellectual disability in the low average and average IQ ranges were more likely to be in the mildly impaired and within normal limits profiles, although exceptions were present. Two individuals with severe intellectual disability were reported to be in the within normal limits profile, while an individual with high average IQ was in the moderately impaired profile. This individual also had a confirmed diagnosis of autism and their parent reported that they required a significant amount of prompting to perform adaptive behaviour tasks.

Strengths and limitations

While this is one of the largest studies to compare the adaptive behaviour profiles of several monogenic disorders, several limitations exist. One of the primary limitations is the biased ascertainment of the conditions included. All cohorts were ascertained for speech and language research. Consequently, individuals referred to the study may have been more likely to have speech and language profiles more severe than might be expected for the wider population of those with the specific syndrome and contributed to lower scores on the Vineland-3, particularly on the communication domain. Having only one measure of behaviour also limited the ability to examine other specific profiles that may (or may not) have emerged in particular syndromes. In future research, inclusion of additional measures of behaviour, executive functioning, communication skills, and co-occurring neurodevelopmental and mental health conditions would provide further insights into the common challenges experienced across NDDs, and whether profiles are like those presented in this study or change when accounting for these other factors.

The use of parent and caregiver reports, and clinician-inferred diagnoses of intellectual disability, limited the ability to examine the effect of intellectual disability on adaptive behaviour profiles for the included conditions. Nonetheless, profiles were consistent with what is expected based on previous literature, with conditions associated with no or milder intellectual disability having stronger adaptive behaviour profiles than conditions associated with more severe intellectual disability. Similarly, diagnoses of autism and ADHD were confirmed through diagnostic reports for a small sample of children, with most children across cohorts not reported by their parents to have these co-occurring conditions. Given that the primary research studies, from which the cohorts were drawn, aimed to characterize the speech and language phenotype, further assessment was not undertaken for those individuals who were not indicated by their parents to have co-occurring conditions. Therefore, analyses relating to the impacts of co-occurring autism and ADHD should be interpreted with caution. It is possible that some individuals who were not reported to have autism or ADHD by their caregiver may indeed meet the diagnostic criteria for these conditions. The complex nature of the syndromes included in the current study may also lead parents not to seek additional diagnoses because of diagnostic overshadowing, difficulty accessing diagnostic services, or there being minimal change in clinical care if their child were to receive a co-occurring diagnosis. In contrast, parent-reported diagnoses may represent traits of these conditions, but not meet the diagnostic criteria. Nonetheless, our findings are in line with previous research, demonstrating greater impacts on adaptive behaviour when a child has co-occurring autism.^{5, 6}

Additionally, smaller samples for some of the monogenic conditions (e.g. the BRPF1, FOXP2, and DDX3X variants) may not be entirely reflective of the population of individuals with these variants. Individuals included in this study may be biased, with parents of children with more complex needs expressing interest in research to better understand their child's strengths and challenges.

Conclusions

Both polygenic and monogenic NDDs are heterogeneous in their clinical presentations. Current approaches to categorizing NDDs do not accurately capture this degree of variability and therefore the specific needs of the individual. Additionally, categorical approaches fail to consider the overlap in symptoms across the NDD spectrum. Transdiagnostic frameworks for identification and intervention across NDDs provide a child-centred approach that can enable targeted support for the most impactful characteristics, rather than diagnostic-driven characteristics.⁴ The present results provide further evidence to support such frameworks, due to the lack of syndrome-specific adaptive profiles and representation of distinct syndrome groups across putative latent groupings. Movement towards needs-based assessments can identify areas for targeted support, building from areas of relative strength for each individual.