Infant Low IQ and autism

Post Reply New Topic

Infants with very low IQ are at risk of developing childhood autism spectrum disorder (ASD), according to a new study in autism research released on Monday by Queens College.

Led by Kristina Denisova, associate professor of psychology and neuroscience at QC, “The importance of low IQ to early diagnosis of autism,” also establishes that both verbal and non-verbal delays suggest the need for early evaluation and intervention. While many researchers accept the importance of verbal delays, a breakthrough finding of Denisova’s study shows that any sign of lower cognitive ability—for example, not transferring toy blocks between two hands during play or not actively looking for a utensil such as a spoon when dropped–-is potentially important.

The study was supported by funding from the Simons Foundation Autism Research Initiative and the National Institute of Mental Health of the National Institutes of Health (NIMH).

Autism researchers already knew that many children with ASD also have low IQ scores and accept that verbal delays are reliable signs of ASD. About 35% of eight-year-olds with ASD in the United States have intellectual disability (ID), meaning an IQ below 70. However, low IQ was not believed to be a major feature of ASD, and no previous prospective study looked specifically at the cognitive abilities of infants from the general population in order to assess their risk for autism.

Denisova found that if an infant has low verbal or non-verbal early cognitive abilities, there is a 40% greater likelihood of developing ASD in childhood. Her findings should pave the way for parents and caregivers of infants to seek a medical evaluation at any sign of atypical cognitive development, and she hopes this will be especially helpful to underserved populations, in particular African American and Hispanic families. Rates of both ASD and ID are especially high in children in the African American community, and often there are delays in age of first diagnosis of their children.

“Low IQ can be considered an early sign of abnormal brain development that leads to autism,” Denisova suggested.

In her new study, Denisova set out to ascertain specifically whether infants having low IQ were at risk of developing ASD by ages two to four, which would allow for earlier and therefore more effective intervention and supports for children and families.

“The average age at ASD diagnosis is relatively late, around four to five years, highlighting the importance of establishing early and reliable ASD markers,” she noted. “Earlier evaluation, diagnosis, and treatment could alleviate the burden for families and society associated with an ASD diagnosis, which is estimated to reach $276–1,011 billion by 2025 in the United States.”

Abstract
Some individuals can flexibly adapt to life's changing demands while others, in particular those with Autism Spectrum Disorder (ASD), find it challenging. The origin of early individual differences in cognitive abilities, the putative tools with which to navigate novel information in life, including in infants later diagnosed with ASD remains unexplored. Moreover, the role of intelligence quotient (IQ) vis-à-vis core features of autism remains debated. We systematically investigate the contribution of early IQ in future autism outcomes in an extremely large, population-based study of 8000 newborns, infants, and toddlers from the US between 2 and 68 months with over 15,000 cross-sectional and longitudinal assessments, and for whom autism outcomes are ascertained or ruled out by about 2–4 years. This population is representative of subjects involved in the National Institutes of Health (NIH)-funded research, mainly on atypical development, in the US. Analyses using predetermined age bins showed that IQ scores are consistently lower in ASD relative to typically developing (TD) children at all ages (p < 0.001), and IQ significantly correlates with social, non-social, and total Calibrated Severity Scores (CSS) on the Autism Diagnostic Observation Schedule (ADOS) (p<0.01). Lower IQ is associated with greater autistic impairments. Note, verbal IQ (VIQ) is no better than the full-scale IQ to predict ASD cases. These findings raise new, compelling questions about potential atypical brain circuitry affecting performance in both verbal and nonverbal abilities and preceding an ASD diagnosis. This study is the first to establish prospectively that low early IQ is a major feature of ASD in early childhood.
Lay Summary

The role of IQ scores in autism remains debated. We systematically investigate the contribution of early IQ in an extremely large study of 8,000 children between 2 and 68 months with autism outcomes by about 2–4 years. We show that IQ scores are consistently lower in ASD relative to TD children. This study is the first to establish prospectively that low early IQ is a predictor for ASD diagnosis in early childhood.

INTRODUCTION
Intellectual ability is currently conceptualized as a distinct dimension relative to the core features of Autism Spectrum Disorder (ASD). However, epidemiological data from different countries indicate a high prevalence of low intelligence quotient (IQ) scores in children with ASD. According to the most recent ASD surveillance estimates in the US, 35% of 8-year-old children with ASD have Intellectual Disability (ID) (IQ < 70, Maenner et al., 2021). Similarly, an epidemiological study of 7–12-year-old children in South Korea reveals that about 1/3 of children with ASD have IQ less than 70 (33% in the high probability group and 25.9% from the general population, Kim et al., 2011). A UK report on a set of children (N = 156, 10–14 years) seen as part of an epidemiological Special Needs and Autism Project (SNAP) finds that as much as 55% of the children with ASD have an intellectual disability (IQ < 70) (Charman et al., 2011).

It is still open to question whether lower IQ might be a secondary effect of severe autism symptoms, or whether on the contrary it might causally affect the severity of the symptoms and if so, for which subset of individuals with ASD. Early case studies hint at the contribution of higher IQ scores to future social adjustment outcomes in children with autism (Rutter et al., 1967, p. 11), with low IQ indicating a poor autism prognosis (Carr, 1976). Higher IQ in preschool as well as communicative speech before 6 years of age was found to be associated with better outcomes and prognosis for children with autism (e.g., Gillberg & Steffenburg, 1987). A recent literature review indicates that positive outcomes are more likely for “individuals with higher childhood IQs and language development by the age of 5–6 years” (Levy & Perry, 2011).

The association between autism, IQ, and adaptive functioning has been probed in recent epidemiological and prospective cohort studies. (Impairment in everyday adaptive functioning is one of the essential criteria for diagnosing ID in DSM-5 (American Psychiatric Association, 2013), p. 37). Children with ASD drawn from the general population have lower adaptive skills relative to their IQ scores; moreover, lower adaptive scores associate with magnitude of “early social impairment” (Charman et al., 2011). Studies with infants at a high familial, genetic risk (HR) for ASD (due to an older sibling diagnosed with ASD) indicate that lower IQ is associated with lower adaptive functioning (Bussu et al., 2019; Zwaigenbaum et al., 2021). Further, a study by Salamone and colleagues (2018) examining subdomain scores on the Mullen starting with 7 months and up to a 7 years at follow-up in the BASIS sample, reveals consistently lower scores for HR relative to LR infants, with increasing difficulties in adaptive behaviors detected in HR-ASD-positive, but not HR-ASD-negative or LR children (Salomone et al., 2018).

Intellectual capacities and skills that support active acquisition of information and enable efficient learning by the developing brain “on the fly” may contribute to both higher IQ scores and lower ASD symptoms (and/or normative social functioning and social communication). Conversely, a number of studies (e.g., Denisova, 2019; Denisova & Zhao, 2017; Jones & Klin, 2013), in particular, of HR infants for ASD, have explored possible disruption of basic social and/or attentional mechanisms, which may deleteriously impact intellectual functioning.

Earlier researchers were particularly interested in the effect of atypical acquisition of language in children with autism. They argued that impaired language processing ability might restrict normal socialization and social maturity (Lockyer & Rutter, 1970). Indeed, language “plays an important part in the growth of intelligence” (Lockyer & Rutter, 1970). During the first year of life, innate competencies provide scaffolding during language learning and acquisition (e.g., Saffran et al., 1996). Processes underlying language delay in children, such as those who are ‘late talkers’, are likely distinct from processes underpinning atypical communication, both verbal and non-verbal. In early life, cognitive processes may help shape the processing of initially undifferentiated information in the ambient environment, for example, in the auditory stream, and help “attune” the developing brain to features specific to one's native language (Denisova, 2019). One possibility is that the robustness of this tuning process may either impede or facilitate development of communication skills. Recent data indicate that initial speech delay in young children may herald additional problems, including global delay (e.g., Zengin-Akkuş et al., 2018), and suggests a potential overlap between language delay and autism (e.g., Jiménez et al., 2021), including in adults with autism (e.g., Armstrong et al., 2017).

Individual differences in early cognitive abilities, the putative tools with which to navigate novel information in life, to adapt flexibly in changing environments, and to judge well (Binet & Simon, 1916), remain unexplored for their potential role as a major contributor to core social and non-social symptoms of ASD, in particular pertaining to impairments in the theory of mind (Velikonja et al., 2019). Relatedly, not only ASD children show the greatest impairments on socio-cognitive tasks, considered precursors to atypical ability to think about others' minds (Frith, 1989), ASD children also show delays on tasks requiring cooperation with other people (Ellis et al., 2020). Atypical information processing has been consistently detected in children and adults with ASD (e.g., using functional Magnetic Resonance Imaging [MRI] and a cognitive interference task; Denisova et al., 2013, 2016). Potential atypicalities in information processing in early childhood may interfere with ability to adjust thinking to new conditions in life (Stern, 1914).

Given that on average diagnosis is relatively late with around 4–5 years of age (e.g., 51 months [Maenner et al., 2021]), identifying early signs of ASD is important for discovering early mechanisms potentially driven by IQ differences. This average might subsume a wide range and if so, hint at two or more populations: some individuals who are diagnosed relatively early (severe ASD symptoms, lower IQ) and some relatively late (milder symptoms, higher IQ).

Improved understanding of the role of early IQ in autism manifestations would allow the possibility to re-conceptualize how mental and neural development unfolds in some infants. This important aim is challenging, as reflected by the complicated requirement for ascertaining comorbid ASD and ID in clinical practice, such that social communication is lower than what would be expected for “general developmental level” (DSM-5 [American Psychiatric Association, 2013, p. 51]). Further, DSM-5 advises delaying diagnoses of ID in young infants and toddlers, prior to a course of intervention (DSM-5 [American Psychiatric Association, 2013, p. 39]). The relative uncoupling of intelligence levels from diagnosis makes it more challenging to generate research-driven insights on early drivers of ASD as a function of IQ differences in very young children.

A recent HR infant siblings study reported cases with missed early diagnoses around 3 years of age (Ozonoff et al., 2018) of children who were not considered to have ASD at that early time point, but who were later ascertained to have ASD. Importantly, these children later diagnosed with ASD had normal or average IQ (Ozonoff et al., 2018). That is, the cases that were missed during earlier assessments were not characterized by early low IQ. This finding again implicates the important role of early low (but not high) IQ for stable early ASD diagnoses, at least in the high risk population.

Encouragingly, in an early study, when corrected for the reliability of the test (0.90) (given a correlation of about 0.30, observed between 6 months and 3 years; Hindley & Owen, 1978), only relatively small changes are to be expected in the median at the second testing. To increase confidence in the results of this study of cognitive ability estimates at early ages, we interrogate the reliability of early IQ estimates directly with the data at hand. Moreover, whether IQ is reliable in children with ASD per se is not fully clear, but as reported in the HR study above, stable ASD diagnoses are characterized by early low IQ (Ozonoff et al., 2018). Conversely, an important question is whether IQ is reliable when estimated at extremely early ages.

Only a handful of studies examined IQ relative to the age of ASD onset. Low cognitive abilities are detected in children from the general population who are diagnosed with ASD relatively early, under 2 years of age (e.g., Chawarska et al., 2007, cf. Shumway & Wetherby, 2009). Lower IQ is detected in ASD relative to typically developing (TD) individuals in a sample with a wide range of age at first visit (~3 to 39 years of age, initial non-verbal IQ > 70), with the differences diminishing in an adults-only (18 years and above at first visit) between-group comparison (Prigge et al., 2021). This analysis suggests that participants below 18 years are contributing to the overall finding of lower IQ. In a sample with an initial age at first visit around 28–69 months (and no group matching on IQ), the full-scale IQ (ELC) on the Mullen is lower in ASD vs. TD individuals (Girard et al., 2021). Given the evidence of lower IQ in children diagnosed with ASD in early childhood, we may further expect early low IQ scores to associate with worse ASD manifestations in children diagnosed relatively early in childhood.


DISCUSSION
This extremely large prospective study establishes that low IQ is an important feature of individuals diagnosed with ASD in early childhood. The core features of autism in this group are not independent of low IQ: before some of the children had ASD diagnoses, as infants, they had significantly lower cognitive abilities. Very early low IQ is a predictor of future ASD diagnosis made in early childhood. Early low IQ does not predispose all children to autism, since we detected a few of the low IQ cases who do not have ASD. However, low IQ can be considered as an early sign of abnormal brain development that leads to autism. These findings are based on the population involved in current autism research supported by the NIH in the US, and have important implications, considered as follows.

In this study, both verbal and non-verbal IQ, as well as the estimate of overall, IQ or g (ELC, full-scale IQ standard score) are significantly lower in the ASD group relative to TD from the start of life, and decline—however slightly—over time. Moreover, lower IQ significantly associates with worse autism symptoms in the ASD group. Importantly, very early (below 1 year of age) low IQ carries a higher risk ratio for future ASD diagnosis made by around 3–4 years. Thus, taken together, these findings strongly suggest an overall cognitive deficit that signals neurodevelopmental problems very early in life (below 1 year of age) for infants who later are diagnosed with an ASD.

Why does low IQ characterize children diagnosed with ASD in early childhood?
An important concern in the field is that the current definition of autism (subsumed under the ASD label) may be “too vague” (Frith, 2021; Mottron, 2021) resulting in a diminished distinction between classic, or “prototypical” autism cases, and other types of neurodevelopmental disorders. Thus, one may ask whether the ASD group represents cases with a non-specific neurodevelopmental disorder, with the ADOS mainly serving as an adjunct to ascertain difficulties in behavioral adaptation. Could results in our study be due to an overrepresentation of young children with low IQ in this ASD population, potentially due to over-inclusive diagnostic criteria?

Despite the inherent challenges and difficulties associated with differential diagnoses, it is important to note that data in this study do not support the possibility of ‘over-inclusive’ diagnoses in the ASD group. That is because cognitive abilities in the ASD group were significantly more impaired not only relative to the TD group, but also relative to the third group of subjects, for whom ASD has been ruled out but TD status not conferred (‘noASDdetected’: Figure S3, Table S6). If the ASD group represented mostly cases with a non-specific neurodevelopmental disorder, then cognitive abilities of this separate, ‘noASDdetected’ group would be statistically similar to the ASD group, but in fact they are not similar (e.g., on the ELC). Not only were ASD group's cognitive abilities significantly worse relative to TD group, which is the main comparison group in this work, but also children with ASD were significantly worse relative to the group of children without ASD from the ‘noASDdetected’ group (who are not ascertained as TD, and for whom another neurodevelopmental disorder was not ruled out).

Future work is required to parse the phenotypic heterogeneity within the ‘noASDdetected’ group with regard to non-ASD impairments vis-à-vis their higher IQ scores relative to the TD children. For example, it would be interesting to understand if the group's somewhat higher IQ relative to TD (while at the same time lower IQ relative to ASD) supported more normative functioning as children's development unfolded, thereby precluding the diagnosis of ASD.

According to the early formulation of autism, individuals with autism represent a subgroup within a low IQ population, and this subgroup may be characterized by a specific cognitive deficit of information processing (Frith, 1989), in particular, the lack of mentalizing capacity (Frith, 1992). Our findings are overall consistent with the idea that low IQ is a co-occurring feature in ASD, but specifically when ASD is diagnosed in childhood. It could well be that within our low IQ ASD childhood-diagnosed group there are cases of classic autism with a more specific cognitive deficit (i.e., weak central coherence; Frith, 1991), or perhaps of a different, developmentally-specific nature, as yet to be determined. What is the nature of a cognitive impairment that is both specific and predictive of future ASD in infants, within the 1st year of life? This exciting and significant question has not yet been asked empirically in infants—and cannot be resolved at present with these data—requiring future work during the infancy period.

Uncovering the basis of low IQ, early diagnosed ASD children
In this study, the question of whether low IQ is a secondary effect of some other problem cannot be answered at present and is left open. Indeed, one needs to be cautious when considering potential underlying mechanisms that may cause or contribute to these early differences in human behaviors. Here, we have a significant puzzle because it is impossible to make conclusions about the causal direction. An early brain dysfunction in children that is causing the core autistic features may directly or indirectly compromise performance on IQ tests in early life, and some factor (e.g., a genetic abnormality) which compromises this brain function may also affect the core features, directly or indirectly.

The early low IQ scores in early diagnosed children in this study may be an indicator of the impaired integrity of nascent neurodevelopmental function or structure. Again, these children may differ in etiology from individuals who receive diagnoses in adolescence or adulthood and whose IQ scores are in the average range. For this reason, it may not be advised to combine these subgroups for genetic or even neuroanatomical, brain imaging studies, since these children might differ in other (biological) respects as well, and this question should be a focus of future work. For instance, the molecular mechanisms of pathophysiology might differ, a possibility with direct implications for brain and behavior and for therapeutic interventions. The early diagnosed individuals (vs. later diagnosed children) may need different clinical management and treatment. It would be important to conduct in-vivo, non-invasive brain imaging studies to specifically compare brain circuitries in the two putative subgroups (early vs. later diagnosed).

Only a handful of studies investigated the brain basis of low IQ in ASD using Magnetic Resonance Imaging (MRI), in part because it is challenging to acquire scans with low IQ or low functioning individuals and children with neurodevelopmental disabilities including ASD, because of poor tolerability for the MRI environment. While challenging, it is indeed possible to use established, children- and family-friendly MRI data acquisition protocols and to acquire good quality, low-motion scans with school-aged children with ASD while awake (Denisova et al., 2013), as well as while awake or asleep in preschool children (Lee et al., 2021), without sedation.

In particular, a recent MRI study of children followed longitudinally since age 2 included participants with megalencephaly (enlarged head size relative to height), with all children ascertained for ASD or typical development (Lee et al., 2021). In children diagnosed with ASD, large brains as indicated by structural MRI in early childhood remain large later, with the findings significantly different relative to TD children, for the megalencephaly group with ASD who also had lower IQ (Lee et al., 2021). In a study of resting-state fMRI of children with ASD and low verbal and non-verbal performance (vs. TD), Gabrielsen et al. (2018) detected reduced brain connectivity in low-IQ ASD vs. higher-IQ ASD and TD children and adolescents (Gabrielsen et al., 2018). Despite the challenges of acquiring MRI data in children, it is very important to include individuals with low IQ in brain imaging studies in order to be able to generalize MRI findings to all individuals with ASD (Lee et al., 2021). To address new questions raised in the current study, new MRI studies are required to examine brain structure and function of infants with different IQ levels during very early childhood from the general population and who are followed up for autism manifestations as they grow.

Strengths and limitations
While we did not detect a sex difference in IQ at 6 months (i.e., combined girls and boys low IQ resulted in an elevated risk ratio for future ASD diagnosis, as well as in a males-only group), the lack of female-specific finding may be due to a small number of female infants available at 6 months. This question should be investigated in future work with a larger sample of female infants. The MSEL has been a common instrument to assess early developmental and cognitive abilities in the ASD field for many ASD research studies, including the early NIH's CPEA and current ACE projects. While this continuity with a single instrument by the field is what enabled the current project's extremely large dataset, the norms have not been updated in several decades. It would be important to collect new data and to update the Mullen norms in order to reflect the population of infants and children currently growing up in the US. However, in the current study, the normed scores on the Mullen were significantly correlated with scores on DAS-II, an instrument with more recently updated norms. Alternatively, the Bayley scales measure permits characterization of domain-based abilities, including for cognition, language, and motor skills (with norms calculated more recently).

All children were enrolled either as at-risk infants (due to parental or other concerns, such as family history) or as age-matched healthy controls in Autism Centers of Excellence or similar research centers and networks, or as participants in studies investigating normative development, in the US. It is possible that in this study, the children who were ascertained as typically developing may not necessarily represent the “normative” population at large, as these families may differ in some ways to those families unable to enroll their children (for example, proximity to an urban center, financial incentive to participate, or incentive to receive feedback on their children's functioning). However, in the current study, the ASD population was precisely ascertained: young children were recruited as part of the rigorously peer-reviewed research funded by the NIH in the US, with DSM-5 diagnoses supplemented with direct observational measures (the ADOS) of each child's behavior.

Importantly, by using direct observation information as part of the ascertainment, this study revealed that children diagnosed with ASD in childhood have significantly lower IQ in infancy, relative to those not only ascertained as being typically developing, normal children, but also relative to another set of children who are not ascertained to have ASD and are not ascertained as TD (and who may have a non-ASD developmental disorder). In new studies, it would further be extremely important to ensure a more diverse and equitable recruitment of families, including those not necessarily within reach of regular recruitment efforts or regular ‘catchment’ areas, to gather an even more generalizable sample that reflects population diversity.

An important area of investigation for future work would be to develop a more precise, theoretically-driven set of phenotypic criteria for autism manifestations in early childhood, while taking into account the likelihood for a neurobiological or genetic deficit at an individual level. Are there phenotypic metamers of early childhood autism, that is, cases that are phenotypically similar but in which different underlying (neuro)biological factors converge to produce autism manifestations in childhood?

CONCLUSION
This extremely large study investigated systematically the role of individual differences in intelligence in early life vis-à-vis future ASD outcomes in childhood, and reflects the population with ASD involved in research studies currently funded by the NIH in the US. The key discovery is that low early IQ is a major feature of ASD that can be assessed objectively, reliably and early. The current findings raise important new questions on ASD diagnosed in early childhood, including on whether early-diagnosed ASD cases may have different etiology and pathophysiology relative to those later diagnosed. Future work is required to investigate the specific and developmentally unique brain-based and genetic mechanisms underlying the finding of low early IQ in ASD. This study is the first to establish prospectively that in children who go on to have ASD in childhood, cognitive abilities are already low starting from early infancy.

ACKNOWLEDGMENTS
I am immeasurably grateful to Uta Frith, mentor assoluta, for precious mentorship with exacting and colossal guidance in developing and pursuing ideas in this work on early low IQ in childhood autism, and importantly for exceptionally kind patience, unwavering support, and enthusiastic, aspirational encouragement. We are grateful to all NDA staff at the NIH for their many expert consultations and help during beginning stages of this study

I appreciate you writing this post

I thought Autism was a life-long thing. So, I don't see how a low IQ can indicate which children are more likely to develop Autism...though it is quite plausible that it might indicate which children are more likely to be diagnosed with Autism (they had Autism the whole time but were diagnosed later).

But the statement "Lower IQ is associated with greater autistic impairments." sounds like a winner to me. If you are Autistic and stranded on the Wrong Planet (the Planet of Neurotypicals) I would think a higher IQ would increase your chances of muddling through. (I am not the only Autistic in Mensa.)