Handbook of Clinical Neurology, Vol. 111 (3rd series) Pediatric Neurology Part I O. Dulac, M. Lassonde, and H.B. Sarnat, Editors © 2013 Elsevier B.V. All rights reserved
The autistic spectrum 1 2
LAURENT MOTTRON1,2* AND MICHELLE DAWSON2 Department of Psychiatry, University of Montreal, Montral, Quebec, Canada
Centre d’Excellence en Troubles Envahissants du Developpement de l’Universit de Montral, Hoˆpital Rivire-des-Prairies, Montral, Quebec, Canada
The autistic spectrum currently encompasses three clinical subtypes: autism per se, Asperger syndrome, and Pervasive Developmental Disorder Not Otherwise Specified (PDDNOS), all under the overarching DSM-IV category of “Pervasive Developmental Disorders.” Its most recognizable core, autism, is a precocious behavioral constellation of social and communication atypicalities associated with apparently restricted interests and repetitive behavior (RIRB), together with an uneven ability profile. Autism is associated with multiple but heterogeneous genetic, functional, and structural variations whose established links with the autistic behavioral and cognitive phenotype are yet minimal. The heterogeneity of causative mechanisms and associated conditions implicated in autism requires a broad expertise in neuroscience. This gives pediatric neurology a central role in the recognition of autism, the assessment of co-occurring conditions, and the treatment of some of its associated difficulties. This chapter will focus on current understanding and diagnosis of autism as well as available support, with an emphasis on neurological aspects of autism during childhood.
SPECTRUM BOUNDARIES AND EPIDEMIOLOGY More than 65 years after its initial recognition by Kanner, autism is still considered a behavioral syndrome. Its diagnosis continues to depend on the recognition of a cluster of overt positive and negative clinical traits (“symptoms”), a situation that requires agreement among experts. Kanner’s view of autism was relatively clear-cut; his initial definition summed the particularities presented by 11 children who were clearly distinguishable from their peers, even from a nonprofessional perspective. None of the children described by Kanner had a
recognizable neurological genetic syndrome in addition to autism. All had strong speech atypicalities and/or delay. None of these children was considered by Kanner to be intellectually disabled, but their intelligence manifested itself in an atypical way. Autism was stable over time, though some developmental transformations were originally described then later elaborated on in Kanner’s follow-up papers. Lastly, autism was thought to be rare: the first attempt at epidemiology based on early attempts at standard criteria revealed a prevalence of only 4/10 000. The situation is much more complex now, particularly with respect to other-than-autism autistic spectrum diagnoses (ASDs). ASDs are now considered commonplace, with considerable variability in apparent conformity to social norms, and in adaptation, intelligence, and cooccurring conditions. Clinicians and researchers now have to cope with the transient and unsatisfactory results of demands for categorical diagnosis required by the development of research, combined with the pressure of social recognition and service allocation. How did we get to this point? One important factor has been the creation in the 1980s of standardized instruments. The first major instrument was a short rating scale (CARS) which should now be regarded at best as a screening instrument. This was followed by lengthier and more complex standardized diagnostic instruments (SDIs), in the form of a parent interview (ADI) and observation scale (ADOS). These more complex instruments provided standardized and fine-tuned symptom definition, assessment procedures, and scoring which corresponds to DSM or ICD diagnostic criteria. Cut-off scores for diagnostic areas were introduced. A global cut-off corresponded to the DSM-IV diagnostic requirement for six signs, including at least two in the social area, one in the communication area, one in the RIRB area, and one additional sign.
*Correspondence to: Professor Laurent Mottron, Hopital Riviere-des-Prairies, 7070 Perras Boulevard Montreal, Quebec H1E 1A4, Canada. Tel: 514 323-7260 ext: 2173, E-mail: [email protected]
264 L. MOTTRON AND M. DAWSON The ADOS and ADI quickly became the diagnostic of a related condition in a group of adolescents of typical gold standard worldwide, first for research purposes, intelligence and without speech or developmental delay. then in clinical settings, and ultimately in service proviThe recognition of the partial overlap between Asperger sion. Autism is now defined by ADI and ADOS, a situsyndrome and Kanner’s autism had the effect of adding ation that presents pros and cons. These SDIs stimulated to the initial autism category a group with nondelayed research into prototypical autism by standardizing the speech development, but showing signs in the social and diagnoses of participants. The modeling of phenotype RIRB areas similar to those encountered in autism. recognition by these diagnostic instruments also had maBut more important, a standardized definition of jor consequences on the current broadening of autism autism (autism defined as the SDIs) contributed to raise criteria. SDIs helped to make the diagnosis of autism the rate of diagnosis among a large number of children possible in individuals scoring in the normal range of who shared more similarities with autism than with any measured intelligence, but also in individuals already other diagnostic class, without fully satisfying the SDI diidentified with another condition, a situation which inagnostic criteria. This resulted in inflating dramatically the creased phenotype heterogeneity. Accordingly, the speccategory of atypical autism (ICD-10) or PDDNOS (DSMificity of SDIs has been demonstrated toward specific IV), sometimes simply called “other ASD.” Conversely, groups of idiopathic intellectual disability or language/ the use of a cut-off for consensual autism had the collatlearning disorders, but not toward other specific groups eral effect of rejecting “subthreshold” children out of the of neurodevelopmental conditions, for which the autism autism category, and consequently, out of the research score may range from marginally to strongly positive focus. Therefore, whereas autism is plausibly the empiribased on these criteria. Such a lack of validation, which cally best-defined DSM-IV condition, the many times mirrors the absence of exclusion criteria in the DSM-IV larger PDDNOS category is one of the least-defined. It definition of autism toward syndromes other than requires the presence of some signs in the relevant diagschizophrenia, has considerably blurred the boundaries nostic areas while still being subthreshold for autism or of autism. This is especially the case for differential diAsperger syndrome. Perhaps unsurprisingly, inter-rater agnoses between ASDs with typical speech level and expert agreement for PDDNOS is less than 50%. neurodevelopmental conditions with limited or no imThe recognition of an inventory of diagnosable conpact on measured intelligence, such as Tourette synditions associated with the autistic spectrum phenotype drome or some presentations of ADHD. (as per the SDIs) has resulted in the addition of another SDIs also had a major impact on increasing the me10 to 15% of individuals now described as presenting dian measured intelligence or developmental level of with syndromic (or secondary) autism, in contrast with the population meeting diagnostic cut-off – and then, idiopathic or primary autism. There are antagonistic conversely, in limiting its upper range. Accordingly, views of the distinction between primary and secondary meeting ADOS or ADI cut-offs is quasi linearly depenautism. One is that if an additional condition is present in dent on measured intelligence, with systematic reaching a small percentage only of the autistic spectrum, then of cut-offs for children whose measured mental age is this condition has in fact produced a phenocopy of priunder 15 to 18 months. These instruments also miss some mary autism, that is, it mimics a phenotype characterizvery intelligent individuals (especially after puberty), ing a certain condition via a completely different who nevertheless test positive for autism criteria accordmechanism. In contrast, autism may represent the final ing to expert judgment. In addition, diagnostic cut-offs common pathway to an indefinite number of mechaon the SDIs have the damaging effect of likening autism nisms, thus making secondary autism an opportunity to a behaviorally unidimensional category, despite geto understand idiopathic autism and creating in effect netic independence of some of its overt behavioral traits a “human model” of autism. In our opinion, the latter and a diagnostic threshold that can be reached by an position puts an excessive burden on a transient and unenormous number of different combinations of phenofounded consensus on phenotypes. As a reflection of typic dimensions. SDIs are, however, indispensable to this ambiguity, current classifications of autistic-like pinpoint a prototypical category, in which both clinical conditions in the presence of another identified neurodeand research purposes are grounded. In particular, SDIs velopmental diagnosis are somewhat arbitrary. Synhave contributed decisively to ascertain the distinction dromes whose phenotypes present a superficial between autism with or without intellectual disability, resemblance with autism but are associated with a speand intellectual disability without autism, as well as to cific cause (e.g., Rett syndrome) are included in PDD impose on clinical settings the notion that individuals but not in the autistic spectrum, despite satisfying scoring in the normal range of measured intelligence DSM-IV criteria for autism. In contrast, a child with could fully satisfy the diagnostic criteria. fragile X who also satisfies these criteria will rather be Another major contribution to evolving views of autism considered autistic on Axis 1 and fragile X on Axis 3. was the description by Wiener psychiatrist Hans Asperger Meanwhile, some autistic-like behavioral trajectories,
THE AUTISTIC SPECTRUM 265 for example those associated with visual impairment, separate status for primary and secondary autism. Also, would instead be considered phenocopies. the reported bimodal occurrence of the age of first seiAs a result of these multiple additions to its original zures (in infancy or in adolescence) as a characteristic of description, autism was eventually surrounded by the autistic epilepsy has to be modulated by the existence of much broader category of ASDs. Variation in prevalence similar age peaks for the first seizures in the general of autism per se has not been dramatic (reported as 2/ population. In sum, uncertainty about the prevalence 1000 in numerous studies since the mid-1990s), given deof autism and its major neurological and demographic liberate increases in public awareness, planned increases correlates is a consequence of current standards of in availability of services (including diagnostic services), clinical phenotype delineation (the SDIs), the absence and a purposeful improvement in case finding. Howof exclusion criteria toward other neurodevelopmental ever, with ASDs, aggressive ascertainment of secondary conditions, and the arbitrary nature of any categorical PDDNOS or atypical autism in children already receivdelimitation of continuous behavioral phenotypic traits. ing services for other diagnosable conditions has resulted in what appears to be very high prevalence (in DIAGNOSTIC ISSUES one study, 11.6/1000), even when Asperger syndrome Although autism is a brain-based condition, pediatric is quasi nondiagnosed. Within the same study sample, neurology is not the main entry point of autistic children using the same SDIs, the same diagnosing clinicians, etc., differences in diagnostic definition, some of them into the healthcare system. Moreover, the first assessvery subtle, can be responsible for a fourfold difference ment of ASDs requires about 2 days of multidisciplinary in prevalence estimates. Throughout existing epidemioinvestigations, including about 5 hours for the use of logical studies, the most variation and imprecision in SDIs, which is beyond what is expected of a child neurolprevalence is due to the PDDNOS (or atypical autism, ogist. Autism tends to be first suspected by parents or or other ASD) category. sometimes by others such as daycare staff. General physicians, pediatricians, child psychiatrists and, increasSeveral characteristics of the autistic phenotype releingly often, specialized ASD clinics confirm the vant for epidemiological studies reveal by their variability the extent to which the current delineation of the diagnosis. Referrals to neurologists are limited to comautistic spectrum phenotype is malleable. Sex ratio plex or unclear clinical situations, including epilepsy as ranges from a state of quasi balance between boys and well as medical and genetic conditions with a neurologgirls in the case of children with secondary autism ical expression and a greater-than-chance association who are considered intellectually disabled, to as many with ASDs or a partial clinical overlap with its criteria. as 10 boys for one girl in the case of children with AsperThe child neurologist may then be the first to detect in a child the cluster of atypical social interaction signs, ger syndrome. As much as a sex ratio in favor of males is verbal or nonverbal communication delays or atypicalan important component of the autistic phenotype, this suggests that primary and secondary autism are of a ities, repetitive behaviors, restricted interests, and/or different nature, or that there is a problem with an uneven competence profile. phenocopies. The proportion of autistics regarded as The diagnosis can be challenging as each DSM criteintellectually disabled has fluctuated greatly through rion may have a considerably different expression across the decades, depending on how autism has been defined development, from toddlers to adolescents. Narrowly deand on how intelligence has been measured. In addition, fined autism is usually not reliably discernible before the middle of the second year. In secondary autism, manifesthe current reported prevalence of epilepsy ranges from tations of the first diagnosis and especially motor delays 5 to 40% of autistic individuals. Like sex ratio and intellectual disability, the prevalence of epilepsy differs conoften precede or mask autistic signs. A decrease in peersiderably according to whether autism is secondary or oriented initiatives and reactions, including joint attenidiopathic. There are currently no studies on the prevation, orientation toward a parent’s voice, as well as smile, lence of epilepsy in samples with measured intelligence babbling and gestures coupling during interactions with in the normal range and from which secondary autism parents and peers, is generally the first observed manifeshas been excluded. Clinical evidence is in the direction tation. Visual fixation on objects that is atypically prolonged or under unusual angles may be present at that of a prevalence of epilepsy in this population that is time but is less obvious or salient to observers. A period not much higher than in the nonautistic population. The often-repeated figure of 20–30% is clearly an overof quasi absence of oral verbal expression typically folestimate, and would if anything characterize only seclows until 3 to 5 years of age, when echolalia, then stereoondary autism, where epilepsy is considerably more typed speech, appear. In less than 10% of autistics, speech frequent, as well as microcephaly and intellectual diswill not develop, but in the majority of cases, speech will ability. This aggregation, along with the aboveflourish, still varying considerably in terms of functional mentioned gender balance, again points toward a value. The maximum number of recognizable signs is
L. MOTTRON AND M. DAWSON
observed around the ages of 3 and 6 to 7, due to the presence by this age of language signs which are characteristic if not definitive, and the general lessening of apparent social atypicalities with age. The notion that autism is a lifelong condition has to be tempered by the absence of predictors of adaptive outcome at toddler age, which varies considerably. SDIs therefore take the 4 to 5 years of age period as the time frame during which retrospective ascertainment has to take place, whatever developmental transformation occurs. The variability of outcomes seems to be directly dependent on two factors: how early the diagnosis was made (the later the diagnosis, the more stable it is), and how close to prototypical autism the child was at this age (the greater the distance from the prototype, the less stable the diagnosis). The developmental trend of strictly defined autism contrasts with that of Asperger syndrome, which may involve precocious and spectacular speech abilities. The most difficult diagnostic issue consists in differentiating the apparent decrease in socio-competence inherent to any neurodevelopmental diagnosis, as well as the restriction in the variety of interests resulting from any developmental delay, from their ASD equivalent. A full psychiatric investigation of differential diagnosis is justified in the presence of the major autistic traits, whether or not a neuropsychiatric co-occurring condition is suspected. The main neurodevelopmental conditions of psychiatric relevance whose phenotype overlaps with the autistic spectrum include Tourette syndrome, ADHD and developmental dysphasia in their strongest forms, childhood depression, and social phobia; with current SDIs there may also be an overlap (“quasi-autism”) with early development arising from extreme deprivation (e.g., Romanian orphans). Early-onset schizophrenia in its prototypical form is not a diagnostic issue, but schizoid personality is currently one of the differential diagnoses of Asperger syndrome. In the presence of incontestable psychiatric diagnosis anterior or alternative to ASD, we suggest applying an economy principle, i.e., answering the following question: can the socio-communicative atypicalities or limitation in the variety of activities be explained more economically by the first diagnosis? For example, an individual presenting with strong Tourette syndrome will display poor social adaptation due to attention deficit, irascibility, and sideeffects of his temper, with the result of limiting the variety of his activities according to what he successfully processes. This will lead him to score positively for PDDNOS criteria, but this second diagnosis will neither provide information relevant to intervention nor contribute to a better understanding of his adaptive difficulties, which will still have to be addressed in a Tourette syndrome framework. A similar position is recommended in the presence of major ADHD or early deprivation.
Conversely, some neurologically relevant conditions are so frequent in ASDs that they should lead to a systematic examination of ASD when they are the entry point for neurological examination. Whereas “pure” developmental dysphasia, despite definitional limitations in communicative abilities, does not impact on nonverbal social interaction, this syndrome is inherent in its narrow sense to a large fraction of autistic individuals. One third of autistic individuals may fall under this additional diagnosis, which nevertheless needs to be distinguished from delayed echolalia and stereotyped speech intrinsic to the autism diagnosis. Therefore, in the presence of other autism diagnostic criteria in the three major areas, developmental dysphasia should not exclude the search for ASDs. In addition, positive scoring for ADHD criteria is the rule in a majority of children with Asperger syndrome, and consequently, it should not be considered an exclusion criterion for this diagnosis. Epilepsy and regression share a certain number of diagnoses and mechanisms in the context of ASDs. Most persistent forms of childhood epilepsy, especially precocious ones and regardless of their cause, produce complex behavioral regressions overlapping with autism criteria. Regression with epilepsy is more dramatic than the loss of the first words, typically numbering less than ten, which signals autism for one quarter of autistic children. It also occurs at an earlier age (West syndrome) or later age (Landau–Klefner syndrome) than is the case with apparent lost speech in autism. A noticeable syndrome constellation is that of tuberous sclerosis, epilepsy, and autism. Tuberous sclerosis is found in 8 to 14% of individuals displaying narrow autism with epilepsy, and is thereby 100 times more frequent among autistics than in the general population. TS is also one of the most important causes of West syndrome, itself associated with dramatic behavioral and cognitive regression. This results in a clinical picture scoring positive on autism criteria, but also with a development of autistic signs earlier than in idiopathic autism, and a measured age equivalent under the threshold for where distinctions between autism and intellectual disability becomes possible, that is, around 18 months. Lastly, partial seizures are difficult to detect clinically in an autistic child, and even if their prevalence may not be higher in autism, they have to be investigated in the presence of dramatic changes or discontinuity in cognitive performance.
NEUROLOGICAL ETIOLOGY AND INVESTIGATIONS Models As a consequence of the phenotypic and causative heterogeneity of ASD, the search for a unique and common neurocognitive mechanism for all individuals sharing ASD criteria is no longer a research purpose.
THE AUTISTIC SPECTRUM 267 Informative similarity between two autistic phenotypes intelligence has been removed. Understanding of human ranges from maximal (e.g., between two idiopathic autisintentions in autism, an overt index of social cognition, tic children sharing the same variant genes and cognitive has been shown in multiple studies to be at least at the profile) to minimal (e.g., between two individuals with typical level. Numerous findings related to complex pursecondary autism differing in their related neurodeveposive actions, problem-solving, and executive functions lopmental condition). There is also the possibility of little yield apparent advantages and apparent disadvantages, or no informative similarity between primary autism and depending on experimental design and task demands, secondary autism phenocopies. We will therefore limit but no overall picture of a deficit. Long-term memory our presentation of current models to idiopathic autism, performance follows measured intelligence, but it is as each type of secondary autism requires a specific also less influenced by semantic or emotional properties model of the relationship between the two diagnoses inof the memorized material than is the case with volved. Even if limiting ourselves to primary autism does nonautistics. not eliminate the heterogeneity of mechanisms at the geTest scores in the range of intellectual disability in aunetic level, it should help preserve a relative homogeneity tism come with several distinct explanations depending at the microstructural, brain functional, and cognitive on the ASD subgroup under study. The consensual preslevels, at least within ASD subgroups. ence of a significant, and probably dominant, fraction of ASD is now considered a variant of typical informaidiopathic autism with a typical level of measured inteltion processing, i.e., the way humans perceive, memoligence indicates that intellectual disability is not a manrize, manipulate and attribute emotional value to datory component of ASDs. Another percentage of available information. This variation consists in the less apparently intellectually disabled individuals actually automatic or mandatory, compared to nonautistics, displays a normal nonverbal IQ level, indicating that difhigher-order processing of incoming information (cateficulties with speech or language, whatever their cause, gorization, purposive integration, and attribution of play a strong role in how the abilities of some autistic emotional value). A more optional, but not in any way people are judged. Yet another fraction of autistic indiabsent, relation among cognitive processes allows for viduals with low measured intelligence probably correa greater than typical role of perception, including in sponds to the long-term effects of impoverished input tasks not considered to be perceptual in nonautistics. due to unavailability of material that they would be able For typical individuals, a major way to attribute value to process well, such as in the case where an autistic with to incoming information consists in automatically indexdelayed or sparse speech is assumed to have very limited ing it with a social value (personal benefit, social abilities or potential. This mechanism can be inferred adequacy, consistency with collective interests). Autistic from analogous situations involving deaf adults who individuals are therefore judged from a typical viewwere exposed neither to oral nor to sign language, as point to differ most visibly from their peers in their autistics seem more dependent than typical individuals social interests, overt emotions, and more generally on access to specific kinds, amounts and arrangements socially oriented behavior. However, there is now of information that may not be easily or at all available. a consensus that autistic information processing differLastly, intellectual disability in autism coincides, in many ences are domain general, encompassing information cases, with secondary autism, and follows the same regarded as nonsocial (e.g., visuospatial processing). explanations as those used for intellectual disability While nonautistic cognition is dominated by an early, per se (e.g., reduced dendrite spines). automatic, and exclusive bias towards social inforSecond-generation brain-based models now deemphamation, the predominance of perception in autistic size the putative action of single cognitive deficits with cognition allows for enhanced processing of environcascading effects, such as amygdala or cerebellum defimental regularities in visual and auditory information. cits, in favor of distributed reorganization among brain Enhanced perceptual abilities, or peaks of ability, are functions and levels. Most structural studies find smaller solidly demonstrated. They extend from enhanced discells and atypical minicolumnar organization in gray matcrimination of low-level psychophysical auditory (pitch) ter, but no evident neurochemical modifications. Toand visual (luminance-defined gratings) information to gether with increased gray and white matter volume, pattern detection (e.g., visual search) and manipulation this indicates that small-scale neural networks are modi(mental rotation, visuospatial construction). Positive fied in autism, although the relationship of these modifiemotions may be apparently less related to social inforcations with perceptual peaks and autistic behavior is as mation, although not impaired per se, as positive emoyet unknown. At a macroscopic, regional level, imaging tions are associated with strong focused interests and studies (PET, SPECT) point to diminished oxygen conthe sharing of these. In contrast, and despite immense sumption in temporal regions. Several functional MRI efforts, no cognitive deficit has unequivocally been studies demonstrate diminished activity in frontal and found in cases of autism, once the effect of measured temporal regions but enhanced activity in the visual
268 L. MOTTRON AND M. DAWSON cortex. This latter pattern of activity cannot be straight2 SD above median head size. In rare situations evoking forwardly equated with a functional deficit, as many Cowden syndrome and a mutation of PTEN genes, fMRI studies find that autistic participants perform tasks macrocephaly is associated with tumors. Autistic macroat least at the same level as their nonautistic comparison cephaly coincides with the first behavioral autistic signs group, despite strong differences in topography of activoccurring around the end of the first year. Some but not ity. Therefore, some reallocation of regional brain funcall studies report its normalization at an adult age, while tions must take place. For example, typical or superior others found macrocephaly in 30% of adult autistic indiperformance in working memory or problem-solving viduals compared to 2% of controls. Other cerebral tasks is obtained in the presence of an inverted balance structures present inconsistent variations which cannot of frontal-occipital activation. Decreased inter- and intrabe interpreted even in a research context. hemispheric connectivity has been supported by most difMedical etiologies with neurological expression are fusion tensor imaging studies. Such reduced functional of very limited application in autism. All medical hypothcrosstalk between brain regions is also consistent with eses about large-scale causes of ASDs can be firmly disthe diminished section of the corpus callosum, a situation carded. The putative role of vaccines and exposure to which is consistently found in ASDs. lead or mercury as widespread causes of autism is no longer a scientific issue. This strong statement should not be confounded with the demonstrated neurological Investigations complications of vaccination and metallic exposure in The role of neurology in autism differs dramatically an extremely limited number of cases. In the latter situwhether it pertains to research, where it is major, or to clination, as for any neurodevelopmental condition, a partial ical settings, where it is less so, given our ignorance of the overlap with an autistic phenotype is possible. As for the primary cause of autism and therefore our inability to hypothesis that gastrointestinal symptoms are more address it at this scientific level. Whatever uncertainty prevalent in ASDs than in the typical population, even remains about the actual relationship between neurologipreliminary data are missing, making this model a priorcal features like epilepsy, epileptic regression, intellectual ity neither in clinical investigations nor even in research. disability, neurogenetic syndromes, and autism, these conA safe guideline in this regard is to examine each individditions occur concurrently in clinical settings and justify ual for medical signs, without being influenced by cursystematic clinical, and if relevant, paraclinical investigarent thinking, but nonetheless without discarding the tions. When ASD is suspected, its diagnosis in a pediatric possibility when one child happens to encompass one neurological context should follow a threefold procedure: of these models of putative medical cause of autism. unfocused search for neurological, genetic or medical The most frequent medical condition which mimics an signs; clinical characterization of developmental level in autistic profile is sensorial impairment, specifically conthe cognitive, language and motor areas; and explicit asgenital or early blindness and Leber amaurosis. Rare certainment of DSM-IV criteria and their interpretation reported medical conditions are mitochondrial synin relation to the previous variables. We do not recommend drome and prenatal exposure to valproic acid, but routhe use of SDIs in a context other than an autism specialtine metabolic testing is not recommended. ized clinic, as they require a high amount of expertise in Inborn errors of metabolism presenting a more or less their administration and in ASDs in order to avoid phenotypic resemblance with autism (usually combined misinterpretations. with intellectual disability) are quite numerous, the most Brain imaging is one of the main focuses of interest in frequently reported being phenylketonuria when ASD research. However, its clinical use is limited, with untreated, creatine deficiency syndromes, disorders of no justification for structural MRI in the absence of purine metabolism – adenylosuccinate lyase deficiency, other symptoms, for example, epilepsy or otherwise indisorder of GABA metabolism – succinic semialdehyde coherent clinical information. Increased overall brain dehydrogenase deficiency, disorders of pyrimidine size, as well as absolute or relative decrease of the antemetabolism – dihydropyrimidine dehydrogenase defirior and posterior parts of the corpus callosum section, is ciency, homocystinuria, histidinemia, biotinidase the most replicated structural characteristic in autism. deficiency, infantile ceroid lipofuscinosis and other lysoContrary to most other neurological markers, these pesomal disorders, and Sanfilippo syndrome. Metabolic culiarities do not aggregate with intellectual disability screening should be used based on suggestive clinical and/or additional diagnoses, which in turn make them findings like modification of alertness or tonus, early an intense focus of research. An increase in brain size and intractable self injury, intellectual disability, dysconsistently spares frontal lobes but affects gray and morphisms, and any medical, particularly digestive, white matter to a similar extent. Brain size increase is assymptoms. Together, all these conditions account for sociated with head size increase, and is usually around less than 1% of early autistic phenotypes.
THE AUTISTIC SPECTRUM Neuropsychological and communication characterization of ASDs is required for diagnosis as well as for education and support, at all ages and developmental levels. For a nonverbal child or a child who has not yet reached the age of language development, psychological testing may be impossible due to lack of compliance or comprehension of task instructions. This may mean adapting test administration procedures, the modeling of answers, or modifying the way information is collected (for example, activities could take place in natural settings and information gathering would be limited to an in-depth interview with caregivers). While testing situations are generally pleasant to autistic children, nonverbal tasks with self-explaining instructions may sometimes be the only way to measure their competence level. The cognitive and communication assessment must answer four major questions: Is the average level of cognitive performance at the normal or the intellectually disabled level? Is the cognitive and communication profile consistent with an ASD subtype? What is the nature of the communication channel actually allowing for exchange of information? Are there any associated cognitive deficits? The interpretation of intelligence level in ASD is challenging in that some tests and activities clearly underestimate cognitive level and display an isolated, low level of performance, lower than that associated with the majority of tasks and representing the child’s baseline level. In contrast, other tasks are performed at a level markedly superior to this baseline, making the decision about which level of performance adequately represents intelligence an especially difficult one. In autism per se, especially in the first years, language-mediated tasks requiring the understanding of verbal instructions are the most difficult to perform. However, vocabulary tasks requiring the child to point to answers or to decode words may be performed at a typical or even superior level. The lowest performance is usually obtained in the “comprehension” subtest of the Wechsler scale, which measures the ability to face a new social situation presented verbally by means of a verbal answer. The highest performance, frequently calculated at 4 to 6 SD above this low performance, is found in the “block design” subtest and in nonverbal fluid intelligence tasks (Raven’s Progressive Matrices). Other Wechsler subtests typically range between comprehension and block design tasks. Whereas low scores in an individual’s profile may predict an overall adaptation to daily demands without accommodation, high scores predict abilities and intelligence level in adapted conditions. This pattern of performance diverges markedly from that observed in the Asperger subgroup, where some verbal tasks (except the comprehension subtest) often present a level of performance superior to what is expected at the child’s chronological age. In contrast, coding and digit span subtests are usually associated with the lowest levels of performance
due to attention difficulties. Asperger children do not display visual-spatial peaks, but interestingly, and in common with autistics, score higher on intelligence when it is measured by Raven’s Matrices as opposed to Wechsler scales. They also present an enigmatic motor clumsiness, not found in autism, which is revealed by major difficulties in writing, cycling, and catching balls. This clumsiness differs from the motor apraxia commonly observed in autism. Neuropsychological and communication assessment should not solely consist of a catalog of deficits. It should search for and reveal successful channels of communication and specific areas of ability and transmit this information to caregivers. In the range of intellectual disability, when using Wechsler scales, one can frequently observe among preschoolers a typical level of understanding of nonverbal or pictorial information, and among school-age children, of printed, written information. Extensive cognitive and language testing should be conducted in order to search for unexpected and unapparent peaks or deficits. Secondary autism and PDDNOS in particular display heterogeneous cognitive profiles, which can be the reverse of the classic autistic profile, with dramatic consequences on academic performance if they are ignored.
GENETIC ETIOLOGY AND INVESTIGATIONS Models Knowledge about genetic influence on autism has reached a paradoxical state of the art, combining strong scientific evidence of high heritability with limited actual determination of genes and modes of transmission involved, and clinical applications restricted to a small subgroup of the phenotype. Uncontested epidemiological evidence of familial aggregation of autistic traits provided the first demonstration of a genetic role in autism. Differences in concordance between monozygotic twins (60% and above) vs. dizygotic twins (around 5%) is considerable, although phenotypic differences between monozygotic autistic twins are often observed. Concordance between siblings ranges from 3 to 6%, with a possible 8% for families who have more than one autistic child, but there is a substantial drop in aggregation between first- and second-degree relatives. Familial aggregation extends beyond narrow autism: concordance for ASD increases considerably when other ASD subgroups, and a fortiori a broader autistic phenotype, are considered. Familial aggregation is greater for probands with higher measured intelligence, which can be explained by the impossibility of differentiating autism from intellectual disability (whose heritability is overall lower than that of autism) in very low mental-age equivalents. A somewhat different and
270 L. MOTTRON AND M. DAWSON weaker aggregation is observed between one autistic proTurner, Cornelia de Lange, Lesch–Nyhan, Smith– band and multiple psychiatric conditions in relatives, Lemli–Opitz, Sanphilippo, Williams, Cohen, Ito, including mood and anxiety disorders, but the clinical Joubert, FG, ARX, XYY, Klinefelter, PKU, 15q, neurousefulness of this notion is limited by its imprecision. fibromatosis, M€obius, and Smith Magenis. The heterogeneity of the autistic phenotype corresponds to the very large and heterogeneous number of TREATMENT AND SUPPORT ISSUES genes implicated in autism studies, most of them being There is no medical treatment for autism, in the sense either nonreplicated or producing only a small effect size. that an inborn and distributed difference in neuronal orThis has first led geneticists to postulate that an interaction of multiple genes, acting with the same strength, ganization and functional allocation of brain resources is would produce by some kind of threshold effect the as yet out of reach of pharmaceutical agents (which emergence of a discernible phenotype. However, current have, however, been proposed via animal models of, syntheses account differently for this heterogeneity and e.g., fragile X). Parental guidance and adaptive emphasize at least two different genetic mechanisms. measures at home and at school, combined with Instead of common genes with a small, predisposing cognitive-behavioral approaches as required for associated conditions, remain the logical way to address most effect acting in association, the genetic heterogeneity challenges. Some Western countries have invested a conof autism would mainly result from a large number of sporadic neo-mutations, including the recently discovsiderable amount of resources in early behavioral interered copy number variations in 10% of simplex families. vention, such as early intensive ABA (applied behavior Many genes thus involved in ASD are implicated in analysis)-based intervention involving one-on-one trainsynaptic transmission, and in neural networks programing for at least 20, and up to 40 or more, hours per week. ing and development. Those mutations – rare genes with Despite their long history and forceful claims by their a causative effect – would occur in critical but multiple supporters, these costly and arduous methods are not supported by good quality research (e.g., randomized possible regions, different for each child, specifically controlled trials conforming to standards well recogin chromosomes 2, 7, 15, and X. They would explain the emerging relationship between an increased possibilnized in intervention and treatment research for the nonity of having an autistic child and an older parental age. autistic population). Ethical issues also remain Some of these mutations would be transmitted to unaddressed. Therefore the actual effects of ABAfemales, more resistant to their penetrance for unknown based autism interventions, their benefits and harms, reasons, and transmitted in a quasi-dominant, mendelian have not been sufficiently established in the short term, way to the more sensitive male offspring. The latter and even less so in the long term. The neurologist should not attempt to normalize disituation would correspond to multiplex families. A spemensions of behaviors that are intrinsically linked to aucial case has to be made for chromosomal rearrangement, found in 3 to 5% of ASD individuals, for which tism, such as trying to impose typical social or other a causative link with the phenotype may or may not be behaviors (equivalent to demanding that blind people possible, and which raises various and contrasted consemake eye contact and read in a typical manner); nor quences for genetic counseling. should positive focused interests be treated as pathological obsessions. In addition, nobody should try to suppress in an autistic child what is tolerated in a typical Investigations child. Sporadic tantrums belong to the typical repertoire Referral to a clinical geneticist is recommended in the of preschoolers, and a less-than-ideal level of activity becase of minor dysmorphism or any sign of a recognizlongs to typical children in the 6 to 9 age group. Repetable neurodevelopmental syndrome. The same is true itive movements should not be targeted in order to in the case of multiplex families or families exhibiting suppress their stigmatizing effect on parents, inasmuch sporadic occurrence of a neurological condition associas they do not represent a burden to the autistic child ated with autism (e.g., TS, fragile X), or when there is a himself. In most cases of primary autism, self-injurious need to refine genetic counseling. The list of genetic conbehavior can be considered a typical autistic expression ditions whose phenotypes may present some overlap of extreme confusion and negative emotions. Since most with ASD is open, but the most frequently cited, with self-injurious behavior is sporadic and without damaging a reported prevalence rate of less than 5% in autism cophysical consequences, it should be addressed as a crisis horts, are tuberous sclerosis and fragile X. Other comsituation and dealt with as such. Contrastingly, lifemonly mentioned conditions, with a prevalence rate of threatening self-injurious behavior is part of the phenoless than 1%, include the following syndromes: Down, type of a number of genetic conditions associated with Rett, Angelman, Prader–Willi, velo-cardio-facial, secondary autism (e.g., Smith–Lemli–Opitz syndrome).
THE AUTISTIC SPECTRUM However, some co-occurring features of the autistic spectrum may be successfully targeted. Indication of physical or psychological suffering should not automatically be interpreted as an autistic symptom and, therefore, left unexplored. Specific language impairments may be addressed with speech therapy, one-to-one interaction, or alternative communication channels (e.g., keyboards), but the majority of efforts should concentrate on managing communication with the child in family and classroom situations. Support for sensory-motor difficulties, including desensitization but also training of underdeveloped motor sectors, is usually provided by an occupational therapist. Rare co-occurring obsessive-compulsive disorder or Tourette syndrome, once distinguished from the repertoire of repetitive autistic movements, can be medicated if necessary, as is generally the case with any co-occurring condition. This also applies to attention deficit, damaging levels of activity (especially verbal), anxiety disorder, and depressive mood in Asperger syndrome. Heavy behavioral problems, such as aggressive behavior or tantrums, intractable wandering or the socially damaging pursuit of restricted interests, are possible but rare. They justify a combined cognitive-behavioral and pharmacological approach. Existing studies have shown that the benefit of atypical antipsychotics in this case is very limited, and may even be less than that associated with a placebo effect. The use of medication with autistics should therefore always follow an in-depth, but unsuccessful investigation of the role of contextual cues in producing a crisis situation, as well as rational attempts to address
the crisis in that way. For example, most apparent temper tantrums in autistic children of preschool age can be alleviated by teaching parents how to decode the communicative attempts of their child, how to provide him with information and materials that he can process well, and how to address him in a way that he can understand. Similarly, depressive mood states in school-age autistic children are often related to bullying, in which case it is the behavior of bullies which needs to be addressed.
FURTHER READING Manzi B, Loizzo AL, Giana G et al. (2008). Autism and metabolic diseases. J Child Neurol 23: 307–314. Minshew NJ, Sweeney JA, Bauman ML et al. (2005). Neurological aspects of autism. In: FR Volkmar, R Paul, A Klin et al. (Eds.), Handbook of Autism and Pervasive Developmental Disorders. Wiley, Hoboken, pp. 473–514. Moss J, Howlin P (2009). Autism spectrum disorders in genetic syndromes: implications for diagnosis, intervention and understanding the wider autism spectrum disorder population. J Intellect Disabil Res 53: 852–873. Mottron L, Dawson M, Soulie`res I et al. (2006). Enhanced perceptual functioning in autism: an update, and eight principles of autistic perception. J Autism Dev Disord 36: 27–43. Samson F, Mottron L, Soulie`res I et al. (2011). Enhanced visual functioning in autism: an ALE meta-analysis. Hum Brain Map http://dx.doi.org/10.1002/hbm.21307 [Epub ahead of print]. Zhao X, Leotta A, Kustanovich V et al. (2007). A unified genetic theory for sporadic and inherited autism. PNAS 104: 12831–12836.