Qualitative Impairment in Social Interactions

The criteria in this domain refer to a qualitative impairment in reciprocal social interactions, not to the absolute lack of social behaviors. Behaviors range from total lack of awareness of another person to the presence of eye contact that is not used to modulate social interactions. The qualitative nature of this and the communication domains were first included in the DSM-IV [1994].

As infants, some autistic children do not lift up their arms or change posture in anticipation of being held. They may or may not cuddle, or even stiffen when held, and often do not look or smile when making a social approach. The characteristic give-and-take in lap play that is seen in typically developing infants and toddlers is often missing. Typically developing infants and toddlers often take great delight in using their newfound “pointer” finger to request or to show; those with ASD usually do not point to request or show/share. Older children often do not point things out or use eye contact to share the pleasure of seeing something with another person, which is called joint attention or social referencing.

Some children do make eye contact, often only in brief glances, but the eye contact is usually not used to get someone’s attention. Others may make inappropriate eye contact by turning someone else’s head to gaze into their eyes. Autistic children may appear to ignore a familiar or unfamiliar person because of a lack of social interest. Some children do make social approaches, although their conversational turn-taking or modulation of eye contact is often grossly impaired. At the opposite extreme of social interactions, some children may make indiscriminate approaches to strangers (e.g., climb into the examiner’s lap before the parent has entered the room, be unaware of psychologic barriers, or be described as a child that continuously and inappropriately “gets in your face”).

Some children with autism indicate little or no interest in other children or adults and prefer to play alone, away from others. Others play with adults nearby or sit on the outskirts of other children’s play and engage in parallel play or simply watch the other children. Some children involve other children in designated, often repetitive play, but often only as “assistants,” without heeding any suggestions from the other children. Some prefer to serve in the passive role in other children’s play, such as the infant in a game of house, and follow others’ directions. Other children may seek out one specific child with whom there is a limited solitary interest that dominates the entire relationship.

Autistic children may also have no age-appropriate friends, and older children often are teased or bullied. A child may “want friends” but usually does not understand the concept of the reciprocity and sharing of interests and ideas inherent in friendship. They may refer to all classmates as friends; one telling example is the child who said, “Oh, I have many, many, 29 friends, but none of them likes me.” Verbal children may have one friend, but the relationship may be limited or may focus only on a similar circumscribed interest, such as a particular computer game. Often, children gravitate to either older peers, in which case they play the role of followers, or to much younger peers, in which case they become the leaders.

Qualitative Impairment in (Verbal and Nonverbal) Communication

The communication deficits seen in the autistic spectrum are far more complex than presumed by simple speech delay, and they are similar to the deficits seen in children with developmental language disorders or specific language impairments [Allen and Rapin, 1992]. Expressive language function across the autistic spectrum ranges from complete mutism to verbal fluency, although fluency is often accompanied by many semantic (i.e., word meaning) and verbal pragmatic (i.e., use of language to communicate) errors. Some mute autistic children do not respond to their names, and often, they are initially presumed to be severely hearing-impaired.

In early infancy, some children with ASD do not babble or use any other communicative vocalizations, and they are described as quiet babies. Some children have absolutely no spoken language when speech should be developing, and they fail to compensate with facial expressions or gestures. A typically developing infant or toddler may pull his or her mother over to a desired object and then will clearly point to the object they request. In contrast, a characteristic behavior of many autistic children is to use another person’s hand mechanically to point to the desired object, an action called hand-over-hand “pointing.” Other “independent” children make no demands or requests of the parents but learn to climb at a young age and acquire the desired object for themselves.

A common feature of verbally fluent children is their inability to initiate or sustain a conversation, which requires two or more parties communicating in a give-and-take fashion on a mutually agreed topic. Although they may be able to respond relatively well to, ask questions of, or talk “at” another person, the reciprocity inherent in a conversation is often difficult for individuals with ASD.

A hallmark of autistic speech is immediate or delayed echolalia. Immediate echolalia refers to immediate noncommunicative repetition of words or phrases – the child is simply repeating exactly what was heard without synthesizing the intrinsic language. This ability is a crucial aspect of normal language development in infants under the age of 2 years, but it becomes pathologic when still present as the sole and predominant expressive language after the age of about 18–24 months.

Delayed echolalia or scripts refers to the use of highly ritualized phrases that have been memorized, such as from videos, television, commercials, or overheard conversations. The origin of this stereotypic language does not necessarily have to be clearly identifiable. Many older autistic children incorporate the scripts in an appropriate conversational context, which can give much of their speech a rehearsed and often more fluent quality relative to the rest of their spoken language. Children also demonstrate difficulties with pronouns or other words that change in meaning with context and they often reverse pronouns or refer to themselves in the third person or by name. Others may use literal idiosyncratic phrases or neologisms. Verbal autistic children may speak in detailed and grammatically correct phrases, which are none the less repetitive, concrete, and pedantic. If a child’s answers to questions seem to “miss the point,” further history and conversation with the child should be elicited because this is also a hallmark of autistic language deficits.

Some autistic children do not appropriately use miniature objects, animals, or dolls in pretend play. Others use the miniatures in a repetitive, mechanical fashion without evidence of representational play. Some highly verbal children may invent a fantasy world that becomes the sole focus of repetitive play. A classic example of the lack of appropriate play is the fluent autistic preschooler who “plays” by repeatedly reciting a soliloquy of the old witch scene verbatim from Beauty and the Beast while manipulating dollhouse characters in sequence precisely according to the script. When given the same miniature figures and dollhouse but instructed to play something other than Beauty and the Beast, this child is incapable of synthesizing any other play scenario.

Restricted, Repetitive, and Stereotypic Patterns of Behaviors, Interests, and Activities

Some verbal autistic children ask the same question repeatedly, regardless of what reply is given, or they engage in highly repetitive, perseverative play. Others are preoccupied with special interests that are highly unusual. For example, many children are fascinated with dinosaurs, but autistic children may amass exhaustive facts about every conceivable type of dinosaur and about which museums house which particular fossils; these children often repeatedly “share” their knowledge with others, regardless of the others’ interest or suggestions to the contrary.

Many autistic children are so preoccupied with “sameness” in their home and school environments or routines that little can be changed without prompting a tantrum or other emotional disturbance. For example, some insist that all home furnishings remain in the same position, that all clothing to be a particular color, or that only one specific set of favored sheets be used on the bed. Others may eat only from a specific plate when sitting in a specific chair in a specific room, which may not necessarily be the kitchen or dining room. Some children may insist on being naked while in the home but insist on wearing shoes to the dinner table. This inflexibility may also pertain to familiar routines, such as taking only a certain route to school, entering the grocery store only by one specific door, or never stopping or turning around after the car starts moving. Many parents may not be aware that they are following certain rituals to avoid the emotional upheaval, or they may be aware but are too embarrassed to volunteer such information. Within this context, some children have distinct behavioral repertoires that they use to sustain sameness, even when not imposed externally. By adulthood, many of these rituals may evolve to more classic obsessive-compulsive symptoms.

Some children have obvious stereotypical movements, such as florid hand-clapping or arm-flapping whenever excited or upset, which is pathologic if it occurs after the age of about 18–24 months. Running aimlessly, rocking, spinning, bruxism (teeth grinding), toe-walking, or other odd postures are commonly seen in autistic children. Others may repetitively tap the back of the hand in a less obtrusive manner, or touch or smell items. In higher-functioning youngsters, the stereotypic movements may become “miniaturized” as they get older into more socially acceptable behaviors, such as pill rolling [Bauman, 1992; Rapin, 1996].

Many children demonstrate the classic behavior of lining up toys, videotapes, or other favored objects, but others may simply collect things for no apparent purpose. Many are preoccupied with repetitive actions, such as opening and closing doors, drawers, to flipping the tops of trash cans, or turning light switches repetitively off and on. Others repetitively flick string, elastic bands, measuring tapes, or electric cords. Younger autistic children love spinning objects or themselves. Others are often particularly fascinated with water, and they especially enjoy transferring water repetitively from one vessel into another.

Asperger’s Disorder

The validity of Asperger’s disorder as an entity separate from high-functioning (verbal) children with ASD remains controversial [Ariella Ritvo et al., 2008; Frith, 2004; Howlin, 2003; Macintosh and Dissanayake, 2004; Sanders, 2009; Schopler, 1996; Witwer and Lecavalier, 2008; Woodbury-Smith and Volkmar, 2009], and Asperger’s disorder will most likely not be included as a separate entity under the ASD umbrella in DSM-V [in press]. Clinically, the diagnosis of Asperger’s disorder is often inappropriately given as an alternative, more acceptable, “A-word” to high-functioning autistic children [Bishop, 1989]. The similarity and overlap of signs and symptoms of Asperger’s disorder with nonverbal learning disabilities (NLD) additionally expand the spectrum of these developmental disorders [Harnadek and Rourke, 1994; Klin et al., 1995; Rourke, 1989]; a recent report, however, demonstrates a lack of difficulty with spatial- or problem-solving tasks – a main principle in the NLD model – in a small cohort of children with Asperger’s disorder [Ryburn et al., 2009].

In sharp contrast to autistic disorder, DSM-IV-TR Asperger’s criteria state that “there are no clinically significant delays in early language (e.g., single words are used by age 2, communicative phrases by age 3)” [2000, p. 81]. Normal or near-normal cognitive function is also the rule, including self-help skills, “adaptive behavior (other than in social interaction), and curiosity about the environment in childhood” [1994, p. 77]. Although absence of language delay is required for diagnosis, the DSM-IV definition of single words by age 2 and communicative phrases by age 3 is none the less considerably outside the recognized norm for language development [Coplan and Gleason, 1993; Rossetti, 1990; Sanders, 2009; Zimmerman et al., 2002]. Asperger’s criteria for the qualitative impairments in social interaction and restrictive and repetitive patterns of behaviors and activities are identical to those for autistic disorder (for a recent review, see Woodbury-Smith and Volkmar [2009]).

High verbal skills are the rule in Asperger’s disorder, which typically leads to later clinical recognition than with autistic disorder [Volkmar and Cohen, 1991; Woodbury-Smith and Volkmar, 2009]. Despite the DSM-IV definition [1994, 2000], language in Asperger’s disorder is clearly not typical or normal. For example, there usually is pedantic and poorly intoned speech, poor nonverbal pragmatic or communication skills, and intense preoccupation with circumscribed topics, such as the weather or railway timetables [Ghaziuddin and Gerstein, 1996; Klin et al., 1995; Wing, 1981]. Individuals with Asperger’s use fewer personal pronouns, temporal expressions, and referential expressions [Colle et al., 2008]. They often exhibit deficits in the semantics and verbal pragmatics of language, resulting in concrete and literal speech; their answers often miss the point. They also demonstrate deficits in general receptive language [Koning and Magill-Evans, 2001; Noterdaeme et al., 2009; Saalasti et al., 2008] and in prosodic comprehension [Jarvinen-Pasley et al., 2008]. Szatmari et al. [1995] further define this disorder by the complete lack of delayed echolalia, pronoun reversal, or neologisms in language production.

Socially, individuals with Asperger’s disorder are usually unable to form true friendships. Because of their naive, inappropriate, one-sided social interactions and lack of empathy, they may be ridiculed by their peers. Often, they cease their attempts to develop friendships because of the cruel ridicule and then remain extremely socially isolated. Fine and gross motor deficits have been described, including clumsy and uncoordinated movements and odd postures [Jansiewicz et al., 2006; Klin et al., 1995; Nishitani et al., 2004; Rinehart et al., 2006; Wing, 1981]. However, frank motor apraxia is an inconsistent finding [Dziuk et al., 2007; Mostofsky et al., 2006].

Autistic Regression and Childhood Disintegrative Disorder

Approximately 22–35 percent of autistic children initially appear to develop normally until at least 12 months of age, followed by loss of language and/or social skills [Baird et al., 2008; Meilleur and Fombonne, 2009; Rogers, 2004; Tuchman and Rapin, 1997; Wiggins et al., 2009]. Loss of language skills has been found to be specific for ASD [Kurita, 1996; Pickles et al., 2009]. Parents usually report that infants were socially responsive, smiled, waved bye-bye, and said some words, but they then suddenly or gradually stopped speaking and seemed to withdraw. In an on-going surveillance program, Wiggins et al. [2009] found that, not surprisingly, children with a known ASD diagnosis had a higher rate of parentally reported regression than those identified with ASD through the retrospective record review (26 percent vs. 17 percent, respectively). Regression occurred at a median age of 24 months; boys were more likely to demonstrate regression than girls, and at earlier ages. Children who experienced regression were diagnosed with ASD much earlier (mean age 4.2 years) than those without regression (mean age 6.2 years) [Shattuck et al., 2009].

One difficulty hindering a better understanding of autistic regression involves the disentangling of age at onset from age at recognition [Chawarska et al., 2007; Volkmar et al., 1985]. Many children thought by parents to be normal in the first 18 months may indeed show signs or symptoms on retrospective evaluation of home movies and videotapes by as early as 12 months of age [Baranek, 1999; Goldberg et al., 2003; Maestro et al., 2005; Osterling and Dawson, 1994, 1999; Ozonoff et al., 2005; Werner and Dawson, 2005]. Goldberg et al. [2008] found significant concordance between parental report and retrospective analysis of home videotapes of regression only in the language domains.

As recently reviewed by Tuchman [2006, 2009], there is considerable controversy surrounding the relation between autistic regression and epilepsy, with regression associated with an epileptiform electroencephalogram (EEG) approximately 20 percent of the time. Studies report both higher [Hrdlicka, 2008; Kobayashi and Murata, 1998] and lower rates [Baird et al., 2008; Tuchman et al., 1991] of epilepsy in regression. The behavioral phenotypes of autistic regression, Landau–Kleffner syndrome (LKS), and continuous spike-wave during slow-wave sleep (CSWS) overlap considerably, and may represent distinct syndromes based on age of regression, degree and type of regression, and frequency of epilepsy and EEG abnormalities [Tuchman, 2009]. Children with LKS and isolated language regression are more likely to have epileptiform EEGs and seizures than those with an autistic regression [McVicar et al., 2005].

Mitochondrial disorders have recently been reported to be associated with autism, and with regression in particular [Poling et al., 2006]. In a cohort of 25 patients with ASD and definite or probable mitochondrial disease by the Modified Walker and Mitochondrial Disease Criteria [Bernier et al., 2002; Wolf and Smeitink, 2002], Weissman found that 56 percent experienced regression of previously acquired skills; 64 percent of the regressions were multiple, and 43 percent had the regression(s) after 3 years of age [Weissman et al., 2008]. Shoffner et al. [2009] found that 61 percent of children with ASD and mitochondial disease experienced a regression, 71 percent associated with and 29 percent without fever.

By DSM-IV definition, childhood disintegrative disorder (CDD) refers to the rare phenomenon of normal early development until at least age 24 months, followed by the loss of language, social, play, or motor skills which culminate most often in symptoms of autism. Previously called Heller’s syndrome, dementia infantalis, or disintegrative psychosis, CDD usually occurs between 36 and 48 months of age but may occur up to age 10 years [American Psychiatric Association, 1994, 2000]. There is, therefore, much overlap between CDD and autistic regression, which has led to significant controversy [Hendry, 2000; Malhotra and Gupta, 2002]. The category of CDD will most likely be retained in DSM-V [in press] under the umbrella of ASD; however, the diagnostic criteria may be changed to reflect the increased understanding of the phenomenon of regression in autism. The lower age limit of CDD will most likely be increased to age 3 years, with autistic regression occurring prior to age 3 years remaining under autism.

CDD is considered rare, with recent epidemiological data suggesting a prevalence estimate of 2 per 100,000 [Fombonne, 2002b, 2009]. It is usually associated with more severe autistic symptoms than is early-onset autism, including profound loss of cognitive skills resulting in mental retardation. There is a 4:1 male predominance and a mean age of onset of 29 ± 16 months; more than 95 percent demonstrate symptoms of speech loss, social disturbances, stereotyped behaviors, resistance to change, anxiety, and deterioration of self-help skills [Kurita et al., 2004a, b; Mouridsen, 2003; Volkmar and Rutter, 1995]. The risk of epilepsy may be as high as 70 percent [Mouridsen et al., 1999]. Children with CDD after age 3 years are more likely to have seizures than those who regress before age 24 months [Klein et al., 2000; Shinnar et al., 2001; Wilson et al., 2003]. Treatment experience in CDD has been generally limited to anticonvulsant therapy for seizures, although Mordekar et al. [2009] recently reported amelioration of behavior, language, and motor regression after corticosteroid treatment in two children with CDD, seizures, and/or epileptiform EEG patterns.

Pervasive Developmental Disorder – Not Otherwise Specified and Atypical Autism

The diagnosis of atypical autism or PDD-NOS is used when clinically significant autistic symptoms are present involving reciprocal social interactions, verbal or nonverbal communication, or stereotyped behavior, interests, and activities, but criteria are not met for a specific diagnostic category under the umbrella of autistic spectrum or pervasive developmental disorders (e.g., a child who does not meet the required 6 of 12 criteria for the diagnosis of autistic disorder) [American Psychiatric Association, 1994, 2000]. Children whose symptoms are atypical or not as severe are coded under this diagnosis. It should be noted that the DSM-IV definition of PDD-NOS required that a child meet only 1 of the 12 criteria in any of the three core domains; in DSM-IV-TR, the definition was changed to require impairment in the development of reciprocal social interaction and either impairment in verbal and nonverbal communication skills or the presence of stereotyped behavior, interests, and activities. It is expected that this diagnostic category will be eliminated in the DSM-V [in press, Box 48-3].

Box 48-3 Proposed Revision to 299.00 in DSM-V: Autism Spectrum Disorder

Must meet criteria 1, 2, and 3:

1. Clinically significant, persistent deficits in social communication and interactions, as manifest by all of the following:

a. Marked deficits in nonverbal and verbal communication used for social interaction

b. Lack of social reciprocity

c. Failure to develop and maintain peer relationships appropriate to developmental level

2. Restricted, repetitive patterns of behavior, interests, and activities, as manifested by at least TWO of the following:

a. Stereotyped motor or verbal behaviors, or unusual sensory behaviors

b. Excessive adherence to routines and ritualized patterns of behavior

c. Restricted, fixated interests

3. Symptoms must be present in early childhood (but may not become fully manifest until social demands exceed limited capacities)

Rationale

New name for category, autism spectrum disorder, which includes autistic disorder (autism), Asperger’s disorder, childhood disintegrative disorder, and pervasive developmental disorder not otherwise specified

Differentiation of autism spectrum disorder from typical development and other “nonspectrum” disorders is done reliably and with validity, while distinctions among disorders have been found to be inconsistent over time, variable across sites, and often associated with severity, language level, or intelligence, rather than features of the disorder

Since autism is defined by a common set of behaviors, it is best represented as a single diagnostic category that is adapted to the individual’s clinical presentation by inclusion of clinical specifiers (e.g., severity, verbal abilities, and others) and associated features (e.g., known genetic disorders, epilepsy, intellectual disability, and others.) A single-spectrum disorder is a better reflection of the state of knowledge about pathology and clinical presentation; previously, the criteria were equivalent to trying to “cleave meatloaf at the joints”

Three domains become two:

1. Social/communication deficits

2. Fixated interests and repetitive behaviors

Deficits in communication and social behaviors are inseparable and more accurately considered as a single set of symptoms with contextual and environmental specificities

Delays in language are not unique nor universal in autism spectrum disorder and are more accurately considered as a factor that influences the clinical symptoms of autism spectrum disorder, rather than defining the autism spectrum disorder diagnosis

Requiring both criteria to be completely fulfilled improves specificity of diagnosis without impairing sensitivity

Providing examples for subdomains for a range of chronological ages and language levels increases sensitivity across severity levels from mild to more severe, while maintaining specificity with just two domains

Decision based on literature review, expert consultations, and workgroup discussions; confirmed by the results of secondary analyses of data from Collaborative Programs of Excellence in Autism (CPEA) and Studies to Advance Autism Research and Treatment (STAART), University of Michigan, Simons Simplex Collection databases

Several social/communication criteria were merged and streamlined to clarify diagnostic requirements

In DSM-IV, multiple criteria assess the same symptom and therefore carry excessive weight in making diagnosis

Merging social and communication domains requires a new approach to criteria

Secondary data analyses were conducted on social/communication symptoms to determine the most sensitive and specific clusters of symptoms and criteria descriptions for a range of ages and language levels

Requiring two symptom manifestations for repetitive behavior and fixated interests improves specificity of the criterion without significant decrements in sensitivity. The necessity for multiple sources of information, including skilled clinical observation and reports from parents/caregivers/teachers, is highlighted by the need to meet a higher proportion of criteria

The presence, via clinical observation and caregiver report, of a history of fixated interests, routines, or rituals and repetitive behaviors considerably increases the stability of autism spectrum diagnoses over time and the differentiation between autism spectrum disorder and other disorders

Reorganization of subdomains increases clarity and continues to provide adequate sensitivity while improving specificity through provision of examples from different age ranges and language levels

Unusual sensory behaviors are explicitly included within a subdomain of stereotyped motor and verbal behaviors, expanding the specification of different behaviors that can be coded within this domain, with examples particularly relevant for younger children

Autism spectrum disorder is a neurodevelopmental disorder and must be present from infancy or early childhood, but may not be detected until later because of minimal social demands and support from parents or caregivers in early years

(American Psychiatric Association. Proposed Revisions to 299.00 Autistic Disorder in DSM-V, 2010. Retrieved 28 February 2010, from http://www.dsm5.org/ProposedRevisions/Pages/proposedrevision.aspx?rid=94#.)

Risk Factors

Animal Models

No single animal model exists for autism, but several animal models that exhibit some of the major features of autism have provided an opportunity to understand the neural substrates of functional impairments. In the macaque monkey, social behavior is mediated by the amygdala, temporal cortex, orbitofrontal cortex, and superior temporal gyrus [Lord et al., 2000]. In another monkey model (rhesus), bilateral removal of the medial temporal lobes leads to abnormal social behavior during maturation that resolves in adults [Bachevalier, 1994; Machado and Bachevalier, 2003]. Symptoms include abnormal social interaction, absence of facial and body expression, and stereotypic behaviors. Social anxiety and fear have been studied in primates [Amaral, 2002] and in rats [Wolterink et al., 2001], and are related to amygdaloid circuitry.

A behavioral syndrome in Lewis rats with analogies to autism is the result of Borna disease virus infection in neonates [Pletnikov et al., 2003; Weissenbock et al., 2000]. Neonatal infection produces specific behavioral abnormalities, with disturbances in sensorineural development, lower startle responsiveness, and abnormal social play. Proprioceptive systems were abnormal that involved use of hind-limbs and balance. Pathologically, Borna disease virus induces regional neuronal loss in the cerebellum. This model provides some insights into mechanisms of pre- and perinatal infection causing damage to the developing brain, but it does not indicate that Borna disease virus is an etiologic agent in autism.

Other animal models have been developed by knocking out different candidate genes [Lijam et al., 1997], by oxytocin and vasopressin administration [Insel et al., 1999], and by exposing embryos to teratogens such as valproic acid [Ingram et al., 2000]. Behavioral studies have focused on social interaction and memory deficits. Because the neurexin 1-alpha gene has been linked to ASD phenotypes, a knockout mouse model has been developed with analogies to at least one core domain of ASD [Etherton et al., 2009].

Neuropathology

Abnormalities in major cortical and subcortical brain structures have been found through postmortem and magnetic resonance imaging (MRI) studies of autistic subjects. Comprehensive examination of nine autistic postmortem brains was carried out by Kemper and Bauman [1994]. They described three major findings: curtailment of normal development in forebrain neurons, which were smaller and more densely distributed than normal; an apparent congenital decrease in the number of Purkinje cells; and age-related changes in cell size and the number of neurons in the diagonal band of Broca, the cerebellar nuclei, and the inferior olive.

These neuropathologic findings may account for many clinical features of autism. Perinatally acquired lesions in limbic system structures could lead to disruption in memory processes involved in the ability to learn new information. In contrast, striatal and cortical areas involved in habitual memory were spared, potentially relating to the need for sameness, narrow interests, and capacity for rote memory. Disruption of cerebellar function may lead to a number of motor and sensory system deficiencies, including mental imagery, anticipatory planning, and timing and integration of sensory and motor information [Kemper and Bauman, 1998]. Support of brain-tissue banking is critical to continued research into the neuropathologic underpinnings of autism [Pickett, 2001].

Another approach to determining the neuropathologic underpinnings of autism has been suggested by Rodier [2002]. She found that children exposed to thalidomide in the first trimester developed autism at an increased rate, supporting the idea that the brain abnormality originates at the time of closure of the neural tube [Rodier et al., 1996]. This finding was corroborated by postmortem examination of a brain from a subject with autism not exposed to thalidomide, but whose mother was an alcoholic, showing near-complete absence of the facial nucleus and superior olive and narrowing of the brainstem between the trapezoid body and inferior olive. This deficit was reproduced in an HOXA1-knockout mouse model and by exposing rat embryos to valproic acid on the day of neural tube closure [Ingram et al., 2000]. The investigators concluded that central nervous system injuries occurring during or just after neural tube closure can lead to selective loss of neurons derived from the basal plate of the rhombencephalon, and this finding may indicate that the initiating injury in some individuals with autism takes place around the time of neural tube closure.

Additional supporting evidence for this theory comes from a Nova Scotia cohort of 61 autistic children [Rodier et al., 1997]. Forty-two percent of these children with autism had posterior rotation of the external ears, compared with 18 percent of controls. They postulated that this could have been associated with a disruption of otic disc formation in the fourth week of embryonic life and that ear anomalies found in some children with autism could possibly be a marker for initiating events in utero.

An important additional neuropathologic finding described by Casanova et al. [2008, 2002] is the finding of abnormalities in the structure of minicolumns in the brain of autistic individuals. Minicolumns consist of 80–100 neurons, and they are believed to be the smallest unit of functional organization in the cortex. In autistic individuals, the minicolumns were described as more numerous but smaller than those of controls and with less space in between. This structural difference may cause the firing of too many processing units at once and prohibit the units from coherently responding to signals. Over-arousal or under-arousal could easily result. This abnormality could also be responsible for the increased incidence of seizures in individuals with autism.

Serotonin

Elevated blood levels of serotonin (5-hydroxytryptamine – 5HT) in autistic subjects, first reported in 1961 [Schain and Freedman, 1961], are caused by an elevation within circulating platelets, an increase in the serotonin transporter concentration, and a decrease in 5HT₂ receptor binding [Cook and Leventhal, 1996]. Elevated concentrations of a low-molecular-weight peptide that increases the uptake of 5HT into platelets [Pedersen et al., 1999] was demonstrated in autistic subjects. Elevation of platelet serotonin levels has been well studied and generally replicated.

It has been hypothesized that, if siblings of children with autism have higher platelet 5HT levels, it may indicate a higher familial risk [Piven et al., 1991b]. A familial pattern of hyperserotonemia was confirmed by Leboyer et al. [1999] in a sample of 62 autistic subjects and 122 first-degree relatives. Levels of whole-blood 5HT did not change with age in subjects with autism but did change in controls. There does not seem to be a correlation between 5HT levels and specific defined phenotypes [Kuperman et al., 1987]. However, Leboyer et al. [1999] suggested that the level of 5HT in blood could be considered an intermediate or surrogate phenotype that could be correlated with DNA polymorphisms in the 5HT transporter gene.

Serotonergic disturbances, including defects in 5HT transporter expression and decreases in plasma tryptophan, may play a role in the pathophysiology of autism [Croonenberghs et al., 2000; Marazziti et al., 2000]. 5HT has an important role in central nervous system development, affecting social behavior, sleep, aggression, anxiety, and affective regulation, and it plays a role in the modulation of synaptogenesis [Chugani, 2002]. It may have a role in the development of neuropathologic abnormalities in the hippocampus, amygdala, and cerebellum [Bauman and Kemper, 1994] and in thalamocortical connections. Depletion of 5HT in critical developmental periods may also cause a decrease in growth of dendritic spines [Yan et al., 1997]. Using positron emission tomography (PET) scans, Chugani et al. [1997] found alterations of 5HT synthesis in the frontal and thalamic pathways that are important for language production and sensory integration.

About one-half of the patients with Smith–Lemli–Opitz syndrome, a malformation syndrome with mental retardation that is caused by an inborn error in cholesterol biosynthesis [Opitz et al., 2002], exhibit autistic behavior. A mouse model of the Smith–Lemli–Opitz syndrome was created by eliminating DHCR7, which encodes 7-dehydrocholesterol reductase, the terminal enzyme required for cholesterol biosynthesis [Waage-Baudet et al., 2003]. These knockout mice have a 300 percent increase in 5HT immunoreactivity in hindbrains compared with control mice, and 5HT-immunoreactive cells are present in unusual locations and represent an increase in the total number of 5HT neurons and fibers. These observations provide a basis for the autistic phenotype seen in Smith–Lemli–Opitz syndrome, and they support the focus of therapeutic interventions aimed at modulation of the serotonergic system.

Vaccines

There has been public concern about a potential relation between autism and vaccines, undermining the confidence of some parents in accepting the recommended vaccinations. Symptoms of autism generally are noticed first in the second year of life, although they may be recognized only in retrospect, and this is when the measles-mumps-rubella vaccine is given. Targets of concern have been the measles-mumps-rubella vaccine itself and thimerosal, the mercury-containing preservative that was used in childhood vaccines until its removal between 1999 and 2002.

Multiple studies concluded that there was no increase in risk of autistic disorder with exposure to the measles-mumps-rubella vaccine [Chen et al., 2004; Dales et al., 2001; Fombonne and Chakrabarti, 2001; Kaye et al., 2001; Madsen and Vestergaard, 2004; Taylor et al., 1999]; see monograph by Offit [2008]. Additional data derived from passive surveillance systems also consistently failed to detect an association [Patja et al., 2000; Peltola et al., 1998; Plesner et al., 2000]. There has been no evidence that the incidence of regressive autism has increased after administration of the measles-mumps-rubella vaccine. The measles-mumps-rubella vaccine has been given at a constant rate since 1979 in the United States, 1982 in Finland, and 1998 in the United Kingdom and Denmark, but rates of autism have steadily risen [Klein and Diehl, 2004]. The Immunization Safety Review Committee of the Institute of Medicine of the National Academies (IOM) [Stratton et al., 2001] concluded that there was consistent evidence of no causal association between administration of the measles-mumps-rubella vaccine and autism.

However, the IOM report did not think there was sufficient evidence to accept or reject a causal association between exposure to thimerosal and autism [McCormick, 2003]. Exposure to thimerosal is a concern because mercury is a known neurotoxin, and there are claims of abnormal levels of mercury in autistic children who respond to chelation therapy. There are no studies involving low doses, and the pharmacokinetics of ethyl mercury in human infants is unknown. Almost all studies available regarding mercury toxicity are related to exposure to methyl rather than ethyl mercury, and the clinical picture of mercury poisoning from any dose, duration, or age of exposure is different from that of autism [Nelson and Bauman, 2003]. Analysis of data from California, Sweden, and Denmark [Stehr-Green et al., 2003] found that, although thimerosal exposure from vaccines was eliminated in Sweden and Denmark by the early 1990s, the incidence of autism was accelerated. This finding was confirmed by a retrospective cohort study [Verstraeten et al., 2003].

A claim has been made that persistent measles virus found in the gastrointestinal tract of ten children with autism and bowel symptoms was causally related to their autism [Wakefield et al., 1998]. Wakefield’s hypothesis was that measles-mumps-rubella vaccine is a risk factor for children with regressive autism. However, there was no clear clinical description of the subjects, there were no controls, and the history and timing of regression was determined years later retrospectively. Most of the other authors have since rescinded the conclusions of this study [Murch et al., 2004], the paper has been officially retracted by the Lancet [Dyer, 2010; Godlee et al., 2011]. Taylor et al. [1999] found 6.6 percent of children with autism had regression and bowel symptoms, but the onset of symptoms was not associated with the measles-mumps-rubella vaccine. Another prospective report [Fombonne and Chakrabarti, 2001] did not find any association in a large, prospective sample. Further strong evidence against the association of autism with persistent measles virus in the gastrointestinal tract or measles-mumps-rubella exposure came from a case control study by Hornig et al. [2008]. Gastrointestinal tissue samples from children with autism, 90 percent of whom showed behavioral regression, and from controls, was examined for presence of measles virus RNA in a blinded study, with results consistent over three laboratory sites [Hornig et al., 2008]. There were no differences between case and control groups in the presence of measles virus RNA, and gastrointestinal symptom and autism onset were unrelated to timing of the measles-mumps-rubella vaccine.

Screening and Diagnostic Instruments

Symptoms of ASDs vary with characteristics of the child, including age, IQ, and temperament. Although all children with ASDs share core features, there is great heterogeneity within this population. Some characteristics, such as repetitive behavior, are more common at certain ages and levels of cognitive functioning. This heterogeneity poses challenges to recognition during typical medical appointments.

The Child Neurology Society and American Academy of Neurology published a “Practice Parameter for the Screening and Diagnosis of Autism” [Filipek et al., 1999, 2000a]. A two-level approach is recommended. Level 1 involves developmental screening as part of routine well-child care, followed by an ASD-specific screen for children identified with developmental delay. Level 2 involves formal diagnostic procedures to determine whether the child has ASD. This level of diagnostic evaluation is conducted by an experienced clinician, and it uses information gleaned from the medical history, neurologic evaluation, and developmental testing to determine the child’s developmental profile.

Instruments for Autistic Spectrum Disorders Screening

“Red flags” for language development can be easily recognized and should engender prompt referral (Box 48-4). Screening instruments for ASDs are in various stages of development. Presented here are those that have reported good psychometric properties (although they all need further development and psychometric evaluation) and that are available for use by clinicians. These instruments are most sensitive when they follow general developmental screening and are applied when a child has already been identified with delayed or atypical development. Some of these instruments are also being validated for use in the general population as a first-step screening, which would be desirable in many settings. However, administration of ASD-specific screening tools after screening for and identification of atypical development is the method with the most sensitivity (i.e., probability of correctly identifying a child with ASD) and specificity (i.e., proportion of non-ASD cases correctly identified).

Box 48-4 Red Flags for Language Development

Prompt referral should be made for any of the following:

No vocalizations by 6 months

No polysyllabic babbling or gestures by 12 months

No spontaneous (not echoed) single words by 16 months

No spontaneous phrases by 24 months

No spontaneous sentences by 36 months

Any loss of babbling, single words, or phrases, including response to name

The Checklist for Autism in Toddlers (CHAT) [Baird et al., 2000, 2001] was developed in Great Britain, and it is most useful in screening children 18–24 months old. It was designed for use in the general population and to be administered in the home by a medical caregiver. The CHAT is intended to be cost-effective and sensitive to early signs of ASD. Parents respond to nine items asking about their child’s social orienting behaviors, such as pretend play, joint attention, and pointing. The clinician completes five items after observing the child. Unfortunately, when administered in the general population, the sensitivity has been disappointing, with high rates of false-negative results. The developers recommend a screen-rescreen procedure (at 18 and again at 19 months), which increases specificity to within acceptable limits [Charman et al., 2002, 2007].

Scambler et al. [2001] conducted a small study in which they demonstrated better sensitivity (65 percent) and specificity (100 percent) when using the CHAT to distinguish children with autism from among a group already identified with developmental problems. They improved on this further by developing their own criteria (i.e., Denver Criteria) for defining risk status. The CHAT worked best with children whose developmental level was within the 18- to 24-month range, despite them being older in chronologic age. The sample size was small, and the Denver Criteria were developed ad hoc. Although the CHAT looks promising for this type of use, more validation is needed. Development of this tool is on-going, and there are attempts at modifications to improve sensitivity and specificity within the general population.

The Modified Checklist for Autism in Toddlers (M-CHAT) [Robins and Dumont-Mathieu, 2006; Robins et al., 2001] relies only on parent report. There is a 23-item questionnaire completed by a parent. If results of the checklist indicate possible risk for autism, there is a follow-up parent interview. The use of this follow-up interview increases the sensitivity and specificity of the instrument, especially when used in a general pediatric population that has not already been screened for developmental disorders [Kleinman et al., 2008].

The Screening Tool for Autism in Two-Year-Olds (STAT) [Stone et al., 2000, 2008] is composed of interactive items administered by a clinician to children 24–35 months old. It is designed to identify children with autism disorders (not the full range of ASDs) from a population of children identified with developmental delays. The STAT quantifies behaviors in four domains: play, requesting, directing attention, and motor imitation. The instrument requires training in administration and scoring, and it takes about 20 minutes to complete. Strong psychometric properties are reported, although they are derived from small samples, and there is need for further development (e.g., evaluation of usefulness with the full range of ASDs) [Stone et al., 2004]. Preliminary results also suggest that the tool is useful in identifying autism in at-risk toddlers down to 14 months of age [Stone et al., 2008]. The investigators suggest that the qualitative information obtained during the screening can be used to inform intervention goals for individual children. In some settings, this may outweigh the inconvenience of the need for training and the relatively longer administration time.

The screening instruments previously described are designed for early diagnosis. Unfortunately, not all children are diagnosed early, and a neurology practice is likely to encounter older children referred for possible autism. The Social Communication Questionnaire [Berument et al., 1999; Rutter et al., 2003a] was developed to identify older, higher-functioning individuals with ASD. It is a 40-item, parent-completed questionnaire for children 4 years old and older. In a large sample of children referred for possible ASD diagnoses, the instrument performed well when differentiating ASDs from other diagnoses. It also demonstrated good sensitivity and specificity in identifying ASD in school-age children with special learning needs [Charman et al., 2007].

Neurologic Examination

Most investigators report that a small proportion of autistic children have frank macrocephaly or megalencephaly. The distribution of head size is clearly shifted upward, with the mean approximately at the 75th percentile [Bailey et al., 1993; Bolton et al., 1994; Courchesne, 2004; Davidovitch et al., 1996; Lainhart, 2003; Lainhart et al., 2006; Woodhouse et al., 1996]. Some investigators have observed that a large head circumference correlates with higher IQ and tends to be a familial trait [Filipek et al., 1992; Miles et al., 2000]. The large head circumference is not usually present at birth, but it may appear in early to middle childhood with increased rates of growth [Lainhart, 1997; Mason-Brothers et al., 1990; Mraz et al., 2009]. Head circumference is generally normal by adolescence and adulthood [Aylward et al., 2002], as is postmortem brain weight by adulthood [Bauman and Kemper, 1997]. Microcephaly is more common in autistic subjects than in controls, and it is associated with abnormal physical morphology, medical disorders, lower IQ, and seizures [Fombonne et al., 1999; Miles et al., 2000].

Abnormalities in the neurologic examination may include hypotonia, which was observed in about 25 percent of 176 autistic children and in 33 percent of 110 nonautistic, developmentally delayed children. Spasticity was found in less than 5 percent of either group (exclusionary criteria for this sample included the presence of lateralizing gross motor findings) [Rapin, 1996]. Motor apraxia was identified in almost 30 percent of autistic children with normal cognitive function; in 75 percent of retarded, autistic children; and in 56 percent of a nonautistic, retarded control group [Dziuk et al., 2007; Mari et al., 2003; Ming et al., 2007; Mostofsky et al., 2006; Rapin, 1996; Smith and Bryson, 2007; Vernazza-Martin et al., 2005; Williams et al., 2004]. Characteristic motor abnormalities include motor stereotypies, which occurred in more than 40 percent of autistic children but in only 13 percent of the nonautistic control group [Rapin, 1996]. Hand or finger mannerisms, body rocking, and unusual posturing are reported in 37–95 percent of individuals, and they often manifest during the preschool years [Lord, 1995; Rapin, 1996; Rogers et al., 2003]. There may be considerable overlap in the repertoire of stereotypic movements that autistic children and normal or mentally retarded children display, but they are more prevalent in children with autism especially, most particularly those who are low-functioning [Singer, 2009].

Clinical Testing

Electroencephalography

There is currently insufficient evidence to recommend for or against the use of routine screening EEGs in autistic patients. A recent review found that epileptiform EEG abnormalities were present in 10.3–72.4 percent of patients and subclinical abnormalities in 6.1–31 percent [Kagan-Kushnir et al., 2005]. The authors concluded that further research is needed about the significance of the EEG abnormalities in the absence of clinical epilepsy (see section on epilepsy p. 14).

Metabolic Testing

The reported co-occurrence of autistic-like symptoms in individuals with inborn errors of metabolism has led to consideration of screening tests as part of the assessment of patients with severe developmental impairment [Steffenburg, 1991]. Although the percentage of children with autism who prove to have an identifiable metabolic disorder has traditionally been considered to be less than 5 percent [Dykens and Volkmar, 1997; Rutter, 1997; Rutter et al., 1994], the recent recognition of the potential association between autism and mitochondrial disorders is bringing this figure into question. Most biochemical analyses are useful only as research tools in the on-going effort to understand the biology of autism, but metabolic testing clearly is indicated by a history of lethargy, cyclic vomiting, early seizures, dysmorphic or coarse features, mental retardation, questionable newborn screening, or birth outside of the United States [Filipek, 2005]. Although on-going research studies may produce evidence to the contrary, at present, selective metabolic testing should continue to be initiated only in the presence of suggestive clinical and physical findings [Curry et al., 1997; Filipek et al., 1999, 2000b].

Neuroimaging Studies

Routine neuroimaging to evaluate a child with autism and macrocephaly is not warranted unless there is evidence of lateralizing signs on neurologic examination. A review of neuroimaging reports for children with autism found a very low prevalence of focal lesions or other abnormalities, none of which turned out consistently to be more than coincidental findings. PET and single-photon emission computed tomography (SPECT) are used only as research tools and are not indicated in the diagnostic evaluation of autism [Filipek, 2005; Filipek et al., 1999, 2000a].

Tests of Unproven Value

Unsupported claims have led a number of parents to subject their children to various tests in the hope of finding a treatable cause of autism. There is inadequate evidence to support routine clinical testing of individuals with autism for trace elements in the hair [Shearer et al., 1982; Wecker et al., 1985], celiac antibodies [Pavone et al., 1997], allergies (particularly food allergies caused by gluten, casein, and Candida and other molds) [Lucarelli et al., 1995], immunologic or neurochemical abnormalities [Cook et al., 1993; Singh et al., 1997; Yuwiler et al., 1992], micronutrients such as vitamin levels [Findling et al., 1997; LaPerchia, 1987; Tolbert et al., 1993], intestinal permeability [D’Eufemia et al., 1996], stool analysis, urinary peptides, thyroid function [Hashimoto et al., 1991], or erythrocyte glutathione peroxidase [Michelson, 1998].

Feeding Disturbances and Gastrointestinal Problems

Feeding habits and food preferences of children with autism typically are unconventional. Bowers [2002] performed an audit of referrals of autistic children to a dietetic service over a 3-month period and found that, despite selective food preferences in 46 percent, the majority of children had intakes that met or exceeded dietary reference values.

Although there have been reports of gastrointestinal complaints in children with autism dating back more than 30 years [Goodwin et al., 1971; Walker-Smith and Andrews, 1972], the gastrointestinal tract only recently has become a significant focus of study [Filipek, 2005]. In one survey of 500 parents of autistic children, almost 50 percent reported loose stools or frequent diarrhea [Lightdale et al., 2001b]. Other studies found that 19–24 percent of children with autism reported gastrointestinal symptoms, with constipation identified in 9 percent [Fombonne and Chakrabarti, 2001], and diarrhea in 17 percent [Molloy and Manning-Courtney, 2003].

In contrast, others have found that the frequency of gastrointestinal symptoms in ASD is not as common as the literature from gastroenterology clinics might suggest [Black et al., 2002; DeFelice et al., 2003; Kuddo and Nelson, 2003; Mouridsen et al., 2009; Peltola et al., 1998; Taylor et al., 2002]. The recent consensus panel [Buie et al., 2010] reviewed the available literature and concluded that evidence-based recommendations for diagnostic evaluation and management of gastrointestinal problems in children with ASD were not yet available, but that individuals with ASD should receive the same standard of care in the diagnostic work-up and treatment of gastrointestinal concerns, as should occur for patients without ASD. Providers should be aware that problem behavior in patients with ASD may be the primary or sole symptom of underlying medical conditions, including gastrointestinal disorders.

In addition, the consensus panel found that available research data do not support the use of a casein-free diet, a gluten-free diet, or combined gluten-free/casein-free (GFCF) diet as a primary treatment for individuals with ASD. To date, only one small randomized, double-blind crossover study has been published. Similar results were found in both the GFCF diet and non-diet-treated groups [Elder, 2008].

Sleep Disturbances

Sleep disturbances, and particularly abnormalities in sleep-wake cycles, have been a recognized feature of autism for over 25 years (see Didde and Sigafoos [2001]; Ivanenko and Johnson [2008]; Richdale [1999]; and Stores and Wiggs [1998] for reviews). The majority of children with autism have sleep problems, often severe, and usually involving extreme sleep latencies, lengthy nighttime awakenings, shortened night sleep and early morning awakenings [Glickman, 2009; Honomichl et al., 2002; Johnson et al., 2009; Johnson and Malow, 2008; Patzold et al., 1998; Richdale and Schreck, 2009; Schreck and Mulick, 2000; Souders et al., 2009]. Children with autism also have more unusual and obligatory bedtime routines: e.g., requiring that parents hold them, lie down with them, or sit beside their bed; that all family members go to bed at the same time; or that curtains or bedclothes be positioned in a certain way. If these routines are not performed exactly, the result is usually a tantrum or other angry outburst. As might be expected, only the autistic children always followed their bedtime routine [Patzold et al., 1998], and the presence of sleep problems was significantly associated with parental stress [Richdale et al., 2000]. Schreck et al. [2004] recently suggested that both the quantity and quality of sleep per night predicted overall autism scores, as measured by the Gilliam Autism Rating Scale (GARS) [Gilliam, 1995], social skills deficits, and stereotypic behaviors. It is unclear whether sleep disorders in children with autism cause daytime maladaptive behaviors, simply allow them to continue, or actually worsen pre-existing problems [Wiggs and Stores, 1996].

Hering et al. [1999] compared the results of parental questionnaires with electronic movement activity recordings (actigraphy) in three groups of children: group 1 consisted of autistic children whose parents reported sleep difficulties; group 2, autistic children whose parents did not report sleep difficulties; and group 3, typically developing children. The initial questionnaires showed that 50 percent of children in group 1 had sleep disorders versus only 20 percent in group 2 and none in group 3. When sleep was quantified using actigraphy, there were no differences in patterns of sleep between groups 1 and 2, except for more early morning awakening in group 1. These findings support the need for objective study methodologies in these samples. Diomedi et al. [1999] compared polysomnograph parameters in adult autistic individuals, who demonstrated a significant reduction of rapid-eye-movement (REM) sleep, increased interspersed wakefulness, and increased number of awakenings with reduction of sleep efficiency, relative to normal controls (see Harvey and Kennedy [2002] for a comprehensive review of polysomnography in autism and other developmental disabilities).

Epilepsy

Epilepsy occurs frequently in children with autism; approximately one-third will develop epilepsy by adulthood, and all seizure types occur [Volkmar and Nelson, 1990]. There is a bimodal distribution of age of onset, with peaks occurring at younger than 5 years and during adolescence [Spence and Schneider, 2009; Tuchman and Rapin, 2002], and with the rate of epilepsy increased among those with mental retardation or underlying medical conditions. The presence of cerebral palsy or focal motor findings also increases risk. Tuchman et al. [1991] found a prevalence of epilepsy of 14 percent among 314 autistic children, but many were younger than adolescent age; in autistic children with normal or near-normal cognitive function, the risk of epilepsy was less than 10 percent by age 10 years. The rate was 27 percent in the presence of severe mental retardation and 67 percent for those with severe mental retardation and a motor deficit. In a subgroup of 160 children with ASD, normal intelligence, no associated cause, and no family history, the cumulative probability of epilepsy was 6 percent, similar to a rate of 8 percent for children with developmental language disorders. The risk of epilepsy in children with Asperger’s disorder and PDD-NOS was less than 10 percent.

Spence and Schneider [2009] recently reviewed the role of epilepsy and epileptiform EEGs in ASD. They found that the rate of epilepsy, even in idiopathic cases of autism with normal IQ, was indeed significantly higher (13–17 percent) [Canitano, 2007; Canitano et al., 2005; Hara, 2007] than the risk in the general population (1–2 percent). Reports have also focused on the presence of subclinical epileptiform abnormalities [Akshoomoff et al., 2007; Baird et al., 2006a; Canitano, 2007; Chez et al., 2006; Gabis et al., 2005; Giannotti et al., 2008; Giovanardi Rossi et al., 2000; Hara, 2007; Hrdlicka et al., 2004; Hughes and Melyn, 2005; Kim et al., 2006; Rossi et al., 1995; Spence and Schneider, 2009]. Kim et al. [2006] found epileptiform abnormalities in the absence of electrographic and clinical seizures in 19 of 32 children referred for prolonged video EEG monitoring for possible seizure activity. Chez et al. [2006] noted epileptiform EEG abnormalities during sleep in 61 percent of almost 900 children with ASD without a history of epilepsy. Other studies, however, report much lower rates [Canitano, 2007; Gabis et al., 2005; Hara, 2007; Parmeggiani et al., 2007], which may be due to differences in sample characteristics and the use of routine EEGs versus prolonged video monitoring. The high rates of isolated epileptiform abnormalities represent a possible objective physiologic finding in ASD, and the treatment of these discharges in the absence of documented seizures is controversial, pending future well-designed longitudinal studies [Spence and Schneider, 2009].

Autism may follow infantile spasms [Riikonen and Amnell, 1981]. Chugani et al. [1996] found that, among children with infantile spasms, those with bitemporal glucose hypometabolism were likely to manifest autism with severe language impairment and mental retardation. In a recent large population-based study, infantile spasms predicted high risk for ASD, but this was, to a large extent, explained by the association of ASD with the symptomatic origin of the seizures [Saemundsen et al., 2008].

Seizures in children with autism should be treated as they would be in children without autism, with even more attention than usual paid to the possible behavioral and cognitive side effects of antiepileptic drugs. Occasionally, it may be difficult to distinguish repetitive and stereotypic behaviors from symptoms of temporal lobe seizures, and inattention from absence seizures may be construed as abnormal, autistic behavior. It is often necessary to use sedation to obtain EEGs for children with autism.

Congenital Blindness

The comorbidity of autism and congenital blindness has received relatively meager attention in the autism research literature. Autistic symptomatology has been anecdotally associated with congenital blindness (CB) for decades; in some studies up to 30 percent of children with CB were also described as being autistic (see reviews by Cass [1998]; Hobson and Bishop [2003]; Hobson et al. [1999]. Rogers and Newhart-Larson [1989] reported a diagnosis of autism in all five boys studied with Leber’s congenital amaurosis. Ek et al. [1998] found that 56 percent of premature babies with retinopathy of prematurity (ROP) had both autistic disorder and mental retardation, and, of those, one-third had coexistent cerebral palsy; in comparison, only 14 percent of those with hereditary retinal disease had autistic disorder. Janson [1993] postulated that, in blind children with ROP, a behavior pattern of unresponsivity and stereotypic object manipulation emerges between 12 and 30 months to distinguish autistic and nonautistic children with CB. Msall et al. [2004] followed children with ROP at ages 5 and 8 years, and found that 23 percent had epilepsy, 39 percent cerebral palsy, and 44 percent learning disabilities. Of the children with no or minimal light perception or totally detached retinas bilaterally, 9 percent were autistic, as compared with only 1 percent of those with more favorable visual status.

Cass et al. [1994] reported that, of an entire sample of over 600 congenitally blind children of differing etiologies, only 17 percent demonstrated no evidence of additional disabilities and were developing normally at age 16 months when first studied. Subsequently, 31 percent had a regression in their development occurring between 16 and 27 months of age; children who regressed tended to have disorders of central nervous system/optic nerve/retina, while children who did not regress had a purely optical cause for their blindness (e.g., congenital cataracts or glaucoma). The more “central” pathophysiology of the blindness in the regression cohort was subsequently confirmed by neuroimaging studies; the children with developmental regression had more central nervous system lesions than those who did not regress [Waugh et al., 1998].

Brown et al. [1997] reported that almost half of their sample with CB met criteria for autism, and that, even in CB without autism, there were significantly more “autistic features” than seen in matched sighted children. Brown et al. [1997] and Hobson et al. [1999] compared congenitally blind (of various etiologies) and sighted autistic children and noted remarkably similar clinical features. The authors’ clinical impressions were that blind autistic children were less severely impaired than sighted autistic children; none was abnormal in listening response, but most were markedly abnormal in body and object use. Mukaddes et al. [2007] examined 257 blind children for autism, and identified 30 (12 percent) who met criteria for DSM-IV Autistic Disorder. Differentiating factors between those CB children with and without autism included severity of blindness, cerebral palsy, and intellectual level, with greater neurological impairments and more severe blindness in those with CB and autism.

Known Genetic and Other Conditions associated with Autism

Known single-gene defects and diagnosed medical conditions account for about 10 percent of cases of autism [Chakrabarti and Fombonne, 2001; Fombonne, 2002a], and there may be subtle but qualitative differences in the symptoms of ASD in specific syndromes. Bolton et al. [1991] and Rutter et al. [1993] found that 8.1 percent of 151 individuals with autism had specific associated medical conditions, including fragile X syndrome, bilateral deafness, cerebral palsy, multiple congenital abnormalities, and identified chromosomal anomalies. About 3.8 percent had other medical concerns that were less likely to be associated with etiologic factors. The possibility of finding any associated medical condition rises with increasing degrees of mental retardation – approaching 50 percent among persons at the severe and profound levels of cognitive dysfunction [Scott, 1994]. In a survey of multiple studies, the fraction of patients with autism who have an associated medical condition of potential etiologic significance was 0–17 percent, with a mean of 6 percent [Fombonne, 2003a, b]. Defined mutations, genetic syndromes, and metabolic diseases account for up to 20 percent of autistic patients [Benvenuto et al., 2009a, b].

These studies investigated the prevalence of medical disorders in populations of individuals with autism. It is important to differentiate that perspective from the prevalence of an autistic phenotype in a population of individuals with the given disorder. Tuberous sclerosis complex (TSC) and fragile X syndrome (FraX) are prime examples to illustrate this concept. In a population of individuals with autism, only a few will have a diagnosis of TSC or FraX; however, in a population of individuals with TSC or FraX, over 50 percent will have a diagnosis of autism. It is therefore important to recognize the autistic phenotype in associated genetic-metabolic and syndromic conditions to optimize intervention plans for these children and adults [Moss and Howlin, 2009].

Epigenetics

Exogenous factors can modify the control of gene expression. Epigenetics refers to the stable and heritable or potentially heritable changes in gene expression that do not involve a change in DNA sequence [Jiang et al., 2004]. Epigenetic mechanisms involve a signal or stimulus that changes gene expression and may help to explain the onset of symptoms in autism after a period of apparently normal development; it may also play a role in how the environment affects phenotypic expression. Chromosomes originating from one parent may have an abnormal epigenotype that leads to modulation of DNA methylation, which causes gene silencing [Bjornsson et al., 2004a, b].

Genetic Counseling

If one child in a family has ASD, the risk for subsequent siblings is elevated [Chakrabarti and Fombonne, 2001; Micali et al., 2004], and is higher if the child with autism is a girl [Sumi et al., 2006]. Rates of mental retardation without autism are not increased in families with ASDs, suggesting that the mental retardation is part of the autistic disorder, not a separate condition. There is no prenatal test to identify autism, but it is important to look for physical or dysmorphic features in the identified autistic child that may suggest a diagnosable genetic condition such as fragile X syndrome [Miles and Hillman, 2000] or tuberous sclerosis complex.

Cytogenetic methods of karyotyping and molecular analyses for fragile X, and the implications of a cytogenetic or molecular diagnosis for other family members, justify their routine inclusion in the diagnostic evaluation of a child with autism [Abdul-Rahman and Hudgins, 2006; Battaglia and Carey, 2006; Challman et al., 2003; Curry et al., 1997; Filipek, 2005; Herman et al., 2007; Marshall et al., 2008]. Current data do not support extensive clinical genetic testing of all children with ASD [Johnson and Myers, 2007]. However, fluorescent in situ hybridization (FISH) studies targeted for specific deletions or duplications – such as is seen in the 15q syndrome, for example – may prove useful in individual cases. Subtelomeric FISH screening in ASD has failed to identify any abnormalities in a total of 225 subjects [Battaglia and Bonaglia, 2006; Keller et al., 2003; Wassink et al., 2007].

Even if the karyotyping appears normal, a microarray comparative genomic hybridization test (aCGH) is advised. aCGH studies identify CNVs, microdeletions, and microduplications that otherwise would go undetected [Christian et al., 2008; Miller et al., 2010]. As such techniques become more available, they should be extended to children without dysmorphic features, as microdeletions and microduplications are common among those with no dysmorphic features as well [Lintas and Persico, 2009]. When there are implications of a cytogenetic or molecular diagnosis for other family members, especially when another pregnancy may be a possibility, their inclusion in the diagnostic evaluation of a child with autism is recommended [American College of Medical Genetics: Policy Statement, 1994]. Prenatal genetic testing is not recommended, because almost all of the recognizable genetic or genomic abnormalities responsible for ASD are not detectable, with the exception of very rare major chromosomal rearrangements.

Even if no genetic condition is revealed, genetic counseling should include a discussion of the risks of recurrence of ASDs in a subsequent child and the lack of ability to diagnose this prenatally [Folstein and Rosen-Sheidley, 2001; Sumi et al., 2006]. The goal of counseling is to provide as much information as possible but to leave decision-making and interpretation of the genetic information to the families.

Additional studies are needed to understand the role of genetics and environmental influences in autism. Questions can be answered only with increasingly large and varied populations and by advances in phenotypic characterization. It is important to encourage families, particularly those with multiple affected members, to participate in clinical research.

Neuroleptic Agents

Neuroleptics that block dopamine receptors, such as haloperidol, thioridazine, and trifluoperazine, were used until the development of drugs that were more effective in blocking serotonin receptors. Haloperidol decreased motor stereotypies, hyperactivity, withdrawal, and negativism in children with autism [Anderson et al., 1989]. Common side effects are sedation and weight gain [Campbell et al., 1997].

The newer atypical neuroleptics include risperidone, clozapine, olanzapine, and quetiapine. They modulate dopamine (D₂) and serotonin (5HT₂) receptors [Buitelaar and Willemsen-Swinkels, 2000]. Risperidone and aripiprazole have been approved by the Federal Drug Administration (FDA) for the treatment of irritabilty (including aggression, self-injurious behavior, temper tantrums, and mood swings) in school-age children and adolescents with autistic disorder. However, these agents have been associated with weight gain, prolactin increases, and sedation in some patients.

In an 8-week multisite, double-blind, randomized trial of risperidone in 101 children with ASD [McCracken et al., 2002], a relatively low dose of 0.5–3.5 mg/d improved irritability, hyperactivity, and stereotypies. Treatment effects were maintained over 16 weeks of treatment, and discontinuation of the medication resulted in return of behavioral symptoms [Aman et al., 2005]. Side effects included weight gain that averaged 6 pounds over 8 weeks, and mild to moderate fatigue in about half of the children. There was a very low incidence of acute extrapyramidal symptoms. These results have been replicated in children [Hellings et al., 2006; Shea et al., 2004] and adults [Hellings et al., 2006]. Luby et al. [2006] conducted a small trial of risperidone with 24 younger children (ages 2.5–6) with ASD. There was a modest but statistically significant improvement on a global autism rating scale compared to the placebo group after 6 months. Significantly greater weight gain and elevated prolactin levels were observed in the treatment group. A double-blind, placebo-controlled trial with 31 autistic adults [McDougle et al., 1998] similarly found decreased irritability, repetitive behavior, and overall behavioral symptoms in the group treated with risperidone.

An 8-week double-blind, randomized, placebo-controlled trial with aripiprazole in 218 children and adolescents with autistic disorder compared three doses of the drug (5, 10, or 15 mg/d) to placebo [Marcus et al., 2009]. All doses demonstrated improvement in irritability, although only the 5 mg/day group reached statistical significance, possibly due to a high placebo response (35 percent). Improvement was reported within 2 weeks, at a point at which all participants were taking the lowest dose. About 10 percent of the participants withdrew from the study because of adverse effects; the most common side effect leading to discontinuation was sedation. Extrapyramidal symptoms were reported in 23.1 percent of the treatment group, compared to 11.8 percent in the placebo group, and more subjects in the treatment groups received medication to treat these symptoms. Aripiprazole was associated with a higher incidence of weight gain than placebo, although no participants discontinued treatment for this reason.

Other atypical neuroleptics studied with similar results include olanzapine and ziprasidone, although they were assessed only in open trials [Malone et al., 2001] or with very small number of participants [Hollander et al., 2006b].

Concern about side effects has led to some comparisons studies. Miral et al. [2008] compared haloperidol and risperidone in 30 children and adolescents with autistic disorder. Both agents were effective in reducing tantrums, aggression, and self-injury. Risperidone was slightly superior to haloperidol in improving disruptive behavior. There was a significant worsening of extrapyramidal symptoms reported in the group treated with haloperidol. Similar levels of weight gain occurred in each group, and there was a greater increase in prolactin in individuals taking risperidone. Remington et al. [2001] compared haloperidol to the serotonin reuptake inhibitor, clomipramine, in 36 children and adults (ages 10–36 years) with ASD. Haloperidol was superior to clomiprimine in reducing irritability and hyperactivity. Neither medication was superior to placebo in reducing stereotypies or inappropriate speech. There was a higher dropout rate in the clomipramine group due to side effects, particularly behavioral activation.

Opiate Antagonists

Several published trials used naltrexone, a long-acting opiate antagonist that can be taken orally. The hypothesis for benefit is that autism is associated with hypersecretion of brain opioids, including beta-endorphins, and that many symptoms of autism are similar to those induced by opiate administration, such as decreased socialization, repetitive stereotypic movements, and motor hyperactivity [Feldman et al., 1999]. Naltrexone in doses of 1.0 mg/kg daily in 23 autistic children decreased restlessness and hyperactivity [Campbell et al., 1993] in a parallel study design. Side effects were mild gastrointestinal symptoms, appetite decrease, and drowsiness. In a randomized, double-blind, crossover design, naltrexone was associated with modest improvement of behavior in 11 of 24 children, but no improvement in learning occurred [Kolmen et al., 1995]. In a separate, similarly designed study that looked specifically at communication skills, no benefit was seen [Feldman et al., 1999]. Studies by Willemsen-Swinkels and colleagues [Willemsen-Swinkels et al., 1995a, b, 1996, 1999] also failed to prove any benefit, and in some children, stereotypic behaviors were increased.

Serotonin Reuptake Inhibitors

Symptoms causing major disruption in autism, such as anxiety and repetitive and ritualized behaviors, can impair learning. Because of the efficacy of serotonin reuptake inhibitors (SRI; e.g., clomipramine, fluoxetine, sertraline, fluvoxamine, and paroxetine) on anxiety and obsessive-compulsive symptoms, and the finding of serotonin system abnormalities in individuals with autism [Chandana et al., 2005], there has been considerable interest in treating disruptive behaviors in autism with these agents. Results of open-label and observational studies have been mixed but encouraging for the reduction of ritualistic behavior, anxiety, and aggression, as well as behavioral rigidity, obsessive-compulsive disorder symptoms, and stereotypies [e.g., Brasic et al., 1994; Cook and Leventhal, 1996; Cook et al., 1992; DeLong et al., 2002; Fatemi et al., 1998; Hollander et al., 2000; Martin et al., 2003; Owley et al., 2005; Sanchez et al., 1995, 1996; Steingard et al., 1997]. However, results of recent double-blind, placebo-controlled, randomized trials have been mixed, and suggest that efficacy of SRIs may be moderated by age, with better responsivity in adults than in children. Additionally, the evidence for possible developmentally sensitive altered regulation of serotonin synthesis in autistic children provides a rationale for giving serotonergic drugs to very young autistic children in order to improve synaptic plasticity during periods of brain development. Controlled clinical trials enrolling very young children are difficult to undertake, but at least one is in process.

A randomized, double-blind, placebo-controlled trial enrolling adults with autism found improvement in one-half of the fluvoxamine-treated patients compared with placebo, with reduction of repetitive thoughts and behavior, maladaptive behavior, language, social relatedness, and aggression [McDougle et al., 1996]. There were few side effects at a mean dose of 270 mg/day. A crossover trial of six adults using the selective SRI fluoxetine demonstrated significant improvement in obsessive behaviors and anxiety, and PET scans demonstrated fluoxetine-elevated metabolic rates in the right frontal lobes [Buchsbaum et al., 2001].

Results of trials in children, however, have been less encouraging. Low-dose liquid fluoxetine (mean dose of 10 mg/day) was superior to placebo on a measure of repetitive behaviors, but not on a measure of clinician-rated global improvement, in 39 children and adolescents with autism in a randomized, crossover trial [Hollander et al., 2005]. Agitation resulted in a dose reduction in 16 percent of the subjects in the fluoxetine group, compared with 5 percent in the placebo group, but this difference was not statistically significant.

A large double-blind, randomized, placebo-controlled trial of citalopram in children and adolescents with ASD has failed to find the predicted effects on repetitive behaviors. A multisite study with 149 children with ASD (ages 5–17 years) and high levels of repetitive behavior evaluated the efficacy of citalopram in reducing those behaviors [King et al., 2009]. In 12 weeks of treatment with a flexible dose schedule, citalopram did not separate from placebo in the primary outcome measures of repetitive behavior and global improvement. The mean citalopram dose was 16.5 (SD = 6.5) mg/d. Citalopram levels and high parent-reported compliance to treatment suggest that this was an adequate trial and the doses were similar to those previously reported to be effective in open trials. Adverse events were significantly more likely to occur in the citalopram-treated group. The most frequently reported side effects were activation, stereotypy, diarrhea, insomnia, and dry skin or pruritus. Two subjects treated with citalopram had seizures. Of note, there was a high placebo response rate (34 percent), which underscores the need for placebo-controlled clinical trials.

In summary, SRIs, selective SRIs, and other medications affecting serotonin and dopamine levels can reduce specific symptoms of autism, such as ritualized behaviors, stereotypies, rigid behaviors, aggression, and anxiety in adults. Large multisite clinical trials with children found that SRIs did not reduce repetitive behavior, or other disruptive behaviors, in children with ASD. While SRIs appear to be generally safe, children may be particularly sensitive to the behavioral activation of these drugs [Buitelaar and Willemsen-Swinkels, 2000]. Seizures have emerged under SRI treatment in clinical trials in a few instances, although this is a seizure-prone population and it is not clear whether the medications were causal. Additionally, there has been concern about reports of an association of fluoxetine and possibly other selective SRIs with suicidal ideation in depressed children, and the FDA has recommended that children on these medications should be very carefully monitored.

Stimulants and Drugs to Treat Hyperactivity

Hyperactivity is an important target symptom that can be potentially improved with psychostimulant medication [Handen et al., 2000; Quintana et al., 1995]. The Research Units on Pediatric Psychopharmacology (RUPP) Autism Network [Posey et al., 2007; Research Units on Pediatric Psychopharmacology Autism Network, 2005a] conducted a randomized, double-blind, placebo-controlled trial of methylphenidate in children with ASDs and high levels of hyperactivity and/or impulsiveness. Doses at the 0.25 and 0.5 mg/kg level were effective in reducing hyperactivity and impulsivity, but less effective in reducing inattention, at 4 weeks and after 8 weeks’ continuation. The response rate was about 35 percent compared to typical response rates of around 70 percent in non-PDD children with ADHD. About 18 percent of subjects withdrew due to side effects, primarily irritability. Other common side effects included decreased appetite and trouble falling asleep. Some children responded best to lower doses of methylphenidate. Observational data was available on a subset of 33 study participants, suggesting that the benefits of methylphenidate extended to some aspects of social interactions, self-regulation, and affect [Jahromi et al., 2009].

Atomoxetine is a nonstimulant medication for ADHD that inhibits the presynaptic norepinephrine transporter. In a placebo-controlled, double-blind crossover study with 16 children with ASD and ADHD symptoms, atomoxetine was more effective than placebo in reducing hyperactivity. The response rate was similar to that reported for methylphenidate in children with ASD, and the mean highest dose was 44.2 (SD = 21.9) mg/d. Side effects, including gastrointestinal symptoms, fatigue, and racing heart were common but transient [Arnold et al., 2006]. Only one participant terminated the study due to intolerance of side effects.

Two small placebo-controlled trials found that clonidine, an adrenergic receptor agonist, had some effect in decreasing irritability and hyperactivity [Fankhauser et al., 1992; Jaselskis et al., 1992] in children and adults with autistic disorder.

There is little evidence for any beneficial effect on hyperactivity using the selective SRIs. Naltrexone, however, has improved hyperactivity [Campbell et al., 1993; Kolmen et al., 1995; Willemsen-Swinkels et al., 1995a, b, 1996, 1999] in small randomized, controlled trials. In clinical trials testing the efficacy for atypical antipsychotics for severe behavior disturbance, hyperactivity was decreased with risperidone [Hellings et al., 2006; McCracken et al., 2002; Research Units on Pediatric Psychopharmacology Autism Network, 2005b; Shea et al., 2004].

Antiepileptic Drugs

Several antiepileptic drugs have been used for behavioral manifestations of autism, particularly for treating intense rapid mood shifts. In open-label studies, levetiracetam [Rugino and Samsock, 2002] and divalproex sodium [Hollander et al., 2001] appeared to be well tolerated and to improve repetitive behavior, impulsivity, and mood stability. A retrospective study of topiramate in children and adolescents with ASDs suggested that the drug reduced misconduct, hyperactivity, and inattention in 8 of 15 patients. Two randomized, placebo-controlled trials, however, have had mixed results. Lamotrigine was not better than placebo on several parent-report and clinician ratings of disruptive behavior and autism symptoms in a double-blind, randomized, controlled trial of 28 children [Belsito et al., 2001]. Children in both groups showed improvement. In a small placebo-controlled, double-blind trial of levetiracetam in 8- to 17-year-olds (n = 20), there was no difference from placebo on measures of behavioral disturbance, repetitive behavior, and autism symptoms [Wasserman et al., 2006]. In a randomized, double-blind, controlled trial with 13 individuals (mean age = 9 years), divalproex sodium was better than placebo in decreasing repetitive behavior [Hollander et al., 2006a]. Patients with the most robust response had repetitive behaviors of the compulsive type, as opposed to stereotypies. This small trial hypothesized a specific target (repetitive behavior), as opposed to global improvement, and the participants demonstrated high levels of that behavior at baseline. The results need replication in a larger trial.

It has been hypothesized that the mechanism of benefit from these medications might be their effect on subclinical seizures, which have been reported to occur in this population. However, there is no evidence and no controlled trials investigating whether treatment with anticonvulsants in children with autism who have epileptiform discharges but no clinical seizures might improve behavioral outcomes in children with autism [Spence and Schneider, 2009].

Cholinesterase Inhibitors

Acetylcholine (ACh) plays a significant role in attention and memory performance. Acetylcholinesterase inhibitors (AChE) slow the breakdown of ACh. This is the presumed mechanism by which cholinesterase inhibitors, such as donepezil, slow decline in memory, attention, and learning in Alzheimer’s disease. Animal studies have suggested that administration of these agents early in development may also enhance learning in a prospective way. Because postmortem studies have found abnormalities of the cholinergic system and its nicotinic receptors in the brains of individuals with autism [Lee et al., 2002; Perry et al., 2001], there is growing interest in the potential benefit of these drugs in ameliorating neurodevelopmental disorders [Yoo et al., 2007].

Preliminary data provides some support for pursuing larger trials. A retrospective examination of effects of donepezil in eight children with autism suggested improvements in irritability and hyperactivity, but memory and attention were not measured [Hardan and Handen, 2002]. A randomized, controlled 6-week trial was conducted with 43 children with ASDs, examining the effects of donepezil (2.5 mg/d) [Chez et al., 2003]. The investigators concluded that the drug improved language and reduced overall autistic features. However, the statistical analyses pooled blinded and nonblinded data, leaving the results susceptible to confounders such as placebo effects. A prospective, open-label study with 13 children and adolescents with autism treated for 12 weeks with galantamine found reductions in irritability and social withdrawal [Nicolson et al., 2006]. The efficacy of cholinesterase inhibitors is yet to be demonstrated in rigorously designed and replicated trials.

In summary, a number of medications, particularly atypical neuroleptics, psychostimulants, and SRIs, can help decrease specific symptoms associated with autism, such as behavioral outbursts, stereotypic or compulsive behaviors, aggression, anxiety, inattention, and oppositional behavior (Table 48-1; also see chapter 49). Reduction in these behaviors can improve quality of life and promote better opportunities for learning. However, there must be careful attention to matching the medication with the targeted behavior, and to possible developmental differences in treatment response. Side effects are common, and must be monitored closely and weighed carefully against the benefits of the drugs. Unfortunately, there is currently no pharmacologic treatment that has been demonstrated to address effectively the deficits in language, cognition, and social understanding that are core features of autism.

Complementary and Alternative Medicine

A significant number of families seek complementary or alternative medicine treatments, few of which have been studied in well-designed trials [Levy and Hyman, 2008]. One survey [Levy and Hyman, 2003] found that 32 percent of 284 children at a Pennsylvania regional autism center were using complementary and alternative medicine. The investigators suggested that it is important to respect the parents’ belief if the complementary medicine is not toxic, but if the treatment is potentially harmful, negotiating a safer replacement practice should be attempted. Examples of the more commonly used treatments include nutritional supplements, melatonin, hormones, glutein- and caseine-free diet, immunoglobulins, secretin, and chelation therapy.

A review [Nye and Brice, 2005] of vitamin B₆ and magnesium concluded that the few studies available were inconclusive and samples sizes were too small for an adequate test of efficacy. A double-blind, placebo-controlled study of pyridoxine and magnesium in 10 patients found no benefit but no significant side effects [Findling et al., 1997].

Dimethylglycine, a nutritional supplement closely related to the inhibitory transmitter glycine, has been proposed [Rimland, 1990] as helpful for autistic children and adults, but it was ineffective in two double-blind, placebo-controlled trials [Bolman and Richmond, 1999; Kern et al., 2001].

Plasma fatty acid levels have been found to be decreased in children with autism when compared with typical controls [Sliwinski et al., 2006; Vancassel et al., 2001], leading to interest in supplementation with omega-3 fatty acids. One small, randomized, double-blind, placebo-controlled pilot study with 13 children reported a nonsignificant trend toward improvement in hyperactivity after 6 weeks of treatment [Amminger et al., 2007]. Mild gastrointestinal side effects were noted.

Melatonin therapy has been used to treat the sleep disturbances in ASD, based on low plasma levels or urinary excretion of melatonin [Kulman et al., 2000; Mulder et al., 2010]. Melke et al. [2008] reported mutations and polymorphisms in the acetylserotonin methyltransferase (ASMT) gene, which encodes the last enzyme of melatonin synthesis, which result in dramatic decreases in ASMT transcripts in blood cell lines. Anecdotal and open-label studies of melatonin therapy have suggested significant improvements in sleep architecture in children with ASD [Andersen et al., 2008; Giannotti et al., 2006; Levy and Hyman, 2008]. Three recent double-blind, placebo-controlled crossover studies were performed in a total of 70 children with ASD using regular (5 mg in Garstang and Wallis [2006]; 3 mg in Wirojanan et al. [2009]) or 5 mg controlled-release melatonin [Wasdell et al., 2008]. Up to 94 percent of children derived significant benefits in sleep without evidence of significant side effects. Larger long-term, double-blind, placebo-controlled crossover studies will be needed to document the efficacy of melatonin across the behavioral subtypes of ASD.

Another popular treatment for ASDs is a gluten-free, casein-free diet. A small randomized, double-blind crossover study was conducted with 15 children with autism on a 12-week GFCF diet [Elder, 2008]. There were no statistically significant differences in autistic symptomatology or urinary peptide levels when participants were on the GFCF diet. The study did, however, demonstrate the feasibility of such a study and the need for an adequately powered trial.

Immunoglobulin has been administered intravenously in open trials [DelGiudice-Asch et al., 1999; Plioplys, 1998] and not found to be useful. A double-blind, placebo-controlled trial of oral human immunoglobulin was conducted with 125 children and adolescents who had autism and persistent gastrointestinal symptoms [Handen et al., 2009]. There was no significant benefit on gastrointestinal symptoms, measures of autistic symptomatology, or behavior disturbances.

Reported dramatic improvement after the administration of secretin as part of endoscopy in three autistic children [Horvath et al., 1998] led to widespread use by parents. Subsequent blinded, randomized trials did not substantiate its efficacy [Carey et al., 2002; Chez et al., 2000; Coniglio et al., 2001; Corbett et al., 2001; Dunn-Geier et al., 2000; Esch and Carr, 2004; Kaminska et al., 2002; Kern et al., 2002, 2004; Levy et al., 2003; Lightdale et al., 2001a; Molloy et al., 2002; Owley et al., 1999; Ratliff-Schaub et al., 2005; Sponheim et al., 2002; Sturmey, 2005; Unis et al., 2002; Williams et al., 2005]. These trials included single- and repeated-dose protocols.

Proponents of chelation therapy suggest that mercury and other heavy metals may be poorly eliminated by children with autism and that it interferes with neurodevelopment via modulation of immune function and other biochemical systems. Despite the lack of scientific evidence of a link between exposure to mercury or other heavy metals and autism, chelation is widely used. One trial [Adams et al., 2009a, b] administered dimercaptosuccinic acid (DMSA) to 69 children with ASD. Those who had a high urinary excretion of toxic metals (n = 49) were then randomized to either DMSA or placebo for an additional six administrations. Some participants in both groups demonstrated improvement of measures of autism symptoms and disruptive behavior, but there were no significant differences between the two groups, and no comparative group that did not receive any treatment. There were no serious adverse effects reported, although a significant increase in excretion of potassium and chromium was noted. No placebo-controlled studies have examined the safety or efficacy of chelation for treating autism, and deaths resulting from hypocalcemia have been reported from the inappropriate use of a chelator, edetate disodium (EDTA) [Brown et al., 2006], including one boy with autism.

It is challenging to perform clinical trials enrolling children with autism, but without well-designed, blinded studies, safe and effective therapies cannot be determined. Problems include standardization of diagnoses, heterogeneity of target problem behaviors, and lack of cooperation of subjects. Many outcome measures are available, including global measures (Clinical Global Impression of Severity) and others targeted to specific symptoms. Whenever feasible, parents should be encouraged to participate in clinical trials to make progress in validating new pharmacologic and behavioral therapies. Information about on-going trials and their locations can be found at www.clinicaltrials.gov.