MAKING SENSE OF AUTISMJOSEPH J. HALLETT, MD, AND VIREN D'SA, MD

Autistic disorder (AD) is a PODs are Rett's disorder, childhood to define with quantitative, objectivedisorder of brain development. Its disintegrative disorder, Asperger's diagnostic criteria and to associate theneuropathogenesis and etiology are disorder and PDD not otherwise phenotype with a biological marker.thought to be heterogeneous and are specified (PDD NOS). POD NOS At present neither is available althoughknow in only a small fraction of cases. was included to allow clinicians to the Autistic Diagnostic ObservationGenes have a complex indirect and, indicate that a child had a behavioral Schedule (ADOS) offers a meaningfulto date, undefined role in AD. This phenotype that was similarto one of the advancement. Diagnostic criteria arechronic neurologic disorder leaves defined PODs but did not meet the full based on a qualitative determination ofaffected individuals dependent on their criteria for that disorder. In the next the severity of behavioral impairments,families for life. revision of the DSM III infantile autism and demonstrating that a child has

was changed to autistic disorder(AD), the DSM IV prescribed numberthe term used in subsequent editions, of impairments. This qualitativeincluding the DSM IV. determination is subject to the

With the explosion of research confounding of observer bias andin the past 25 years, the concept marked rater-to-rater variance. Thishas emerged that the behavioral confounding variance has been acharacteristics observed in children recurring source of considerablewith AD are not unique to AD but disagreement among physicians,are distributed in lesser forms and in a disagreement that no doubt hasdifferent combinations throughout a perplexed the PCP who is caring for

the child and supporting the family.The ADOS offers an observationalinstrument to reduce the influence of

these confounding variables.

TERMINOLOGY

Autism, autistic disorder and

autistic spec'trum disorder needdefinitions. They are best understoodfrom a historical perspective.

In 1943 Leo Kanner, a child

psychiatrist, described 11 children,most from prosperous and well-educated parents (several werephysicians) with "autistic disturbancesof affected contact.'" He borrowed the

word autistic from the schizophrenialiterature, where it denoted the self-

centered thinking, detached behaviorsand resulting isolation observed inschizophrenia. Following his report,the term infantile autism emerged.

Kanner's report received littleattention in part because the behavioralphenotype was already subsumedunder the diagnosis of childhoodschizophrenia. However, in the 1950sthe psychodynamic theorists resurrectedhis work and hypothesized that childrenbecame autistic because they were notgiven appropriate emotional nurturingby their "refrigerator" mothers. 2.3Thisapproach to autism was the antithesisof Kanner's speculation that infantileautism had a biologic cause. Infantileautism was formalized as a disorderin 1980 when it was included in the

Diagnostic Statistic Manual, thirdedition (DMS 111).4It was categorizedas a pervasive developmental disorder(PPD), a group of behaviorallydefined syndromes that share thecharacteristic of severe and pervasivedisruption of development in "..socialinteraction skills, communication skills

or the presence of stereotyped behavior,interests and activities."4 The other

"AD AND THE

OTHER PDDs ARE

BEHAVIORALLY

DEFINED SYNDROMES

THAT MEDICINE

AND THE SCIENCES

ARE STRUGGLING

TO DEFINE WITH

QUANTITATIVE,

OBJECTIVE

DIAGNOSTIC

CRITERIA ..."

population; AD behaviors representa portion of this distribution. In the1990s the descriptive term, autisticspectrum disorder (ASD), was coinedin the experimental literature. Theterm migrated into clinical literaturewhere it is often used as a diagnosticterm even though a common clinicaldefinition has not been agreed upon.

PERVASIVE DEVELOPMENTALDISORDERS

AD and the other PODs are

behaviorally defined syndromes thatmedicine and the sciences are struggling

AUTISTIC DISORDER

AD is a behaviorally definedsyndrome characterized by severeimpairments in social interaction, thesocial components of language andinflexible and restricted behaviors or

interests. A diagnosis is made basedon a behavior phenotype defined inDSM IV.' Since the diagnosis is basedon behaviors, the AD phenotype canoverlap with other disorders. Forexample, a large percent of childrenwith tuberous sclerosis have the AD

phenotype. Similarly, approximately1% of children with AD have

tuberous sclerosis. This overlap ofphenotype is also seen with congenitalbrain malformations and untreated

phenylketonuria. However, most casesof AD are idiopathic.

AUTISTIC SPECTRUM DISORDERThe basic behavioral traits that

comprise autism have not changedsince Kanner reported them. However,the interpretation of these traits haschanged and so has the type of childreceiving an autism diagnosis. The

1VOL. 88 NO.5 MAY 2005

broadest interpretation is representedby the term autistic spectrum disorder(ASD). It assumes that autism is nota discreet disorder but should be

viewed as a spectrum of conditionsor behaviors unified by impairmentsin social interactions, communicationand restricted behaviors or interests.

Since ASD is defined by behaviorsand not by pathogenesis or causativeagent, a spectrum can include theKanner-type AD in which there issevere and profound impairments aswell as mental retardation, to milder

cases such a POD NOS, to atypicalcases such as high functioning AD(AD with IQ> 70), to cases with noimpairment of language developmentand only mild impairments in socialinteraction and inflexibility suchas in Asperger disorder, to childrenwhose impairments in the past wereconsidered below diagnostic thresholdfor POD. The dilemma present bythis approach is how to distinguishbetween ASD behavior and behavior

that is eccentric, idiosyncratic or oddwhich are in general considered withinthe range of normal.

Support for interpreting autismas ASD comes from studies that findAD behavioral traits, albeit with

varying severity, distributed throughfamilies of children with AD and in

the general population.6. 7.8 Furthersupport comes from functional brainimaging studies (fMRI) that foundsimilar activation patterns in adultswith high functioning autism andAsperger syndrome.9 On the otherhand, the term spectrum is most oftenused in the context of disorders that are

clinically distinct and share a commonetiology while varying in the severity ofsymptoms. The validity of groupingbehaviorally defined phenotypes ofpotentially different etiologies has beenquestioned. AD and the disorderssubsumed under ASD are postulatedto have heterogeneous etiologies whichopen the possibility that the autisticbehavioral phenotype may representa final common pathway for a varietyof disorders. Determining whetheran autistic phenotype represents thefinal common expression of distinctdisorders or represents a spectrum of

disorders with shared neuropathologicalparameters is a question awaitingclarifYing studies.Although the utility of ASD is debated,it has become a popular clinical term,used to define a behavioral phenotypewithout reference to pathogenesis oretiology. An understanding of thisis important because the use of ASDmust be accompanied by a vigorousdiagnostic evaluation for treatabledisorders.

EPIDEMIOLOGY

The epidemiology of AD hasnot changed substantially from thefindings reported in the 1980s andearly 1990s. It is most often diagnosedin males (male to female ratio of 4 to

1). Cases do not segregate by race,level of education, socioeconomic

status or geography. Behavioral signstypically appear before the age ofthree.1o Delays in language (eitherabsence or slow progression) are themost common presenting complaint. II

A small portion have regression aftertypical language development. Sixtyto seventy-five percent of the childrenwith AD have mental retardation.12.13

Cognitive deficits may be moresevere in girls than boys althoughapproximately 40% of boys will havesevere to profound mental retardation.

Organic, behavioral and emotionalcomorbidities are common. (Table 1)

AD has implications for theentire family. Mild language, socialor psychiatric problems can be presentin parents. 14.15 Siblings have a greater,albeit small, risk of AD, languagedisorder, learning disabilities, socialproblems and psychiatric disorders.

PREVALENCE

A rise in the prevalence ofchildren with an autistic diagnosis hasbeen established although a specificprevalence rate has not.12.16.17Populationstudies in the 1980s and early 1990s

found a prevalence ranging from 2 to10 per 10,000 chiidren.13.17After 1994prevalence increased annually withrecent estimates reaching as high as 67per 10,000 for ASD and 4 per 10,000for ADY Since 1994 Rhode Island

has mirrored this trend in its specialeducation population.18 In 2003, 605students with an autistic diagnosiswere educated in Rhode Island schools.

Understanding the factors behind thisrise in prevalence is critical to ourunderstanding of AD and ASD.

Population-based incidence studies,using contemporary diagnostic criteria,are believed to more accurately reflectpotential changes in the occurrence ofdisorders across time periods. One such

Table 1. Comorbidities in autistic disorder

Mental retardation (60-75%)Epilepsy (5% in children, 30% in adults)Phenylketonuria if untreated (5%)Tuberous sclerosis «1 %)NeurofibromatosisCongenital malformationsCerebral palsyDown syndromeHearing impairmentsVision impairmentsLearning disabilitiesSleep disorderSelf-injurious behaviorsAggressive, angry or combative behaviorsOppositional behaviorHyperkinetic behaviorDepressionAnxietyObsessive compulsive disorderTics

DICINE AND HEAL.TH I RHODE ISL.AND

study in Olmsted County Minnesota(OC) found increases in the incidence

of AD from 0.55 per 10,000 childrenin 1980-1983 to 4.49 per 10,000children in 1995-1997.17 A risingincidence among young children wasresponsible for most of the increased.During the same period the incidenceamong children older than 10 years ofage remained stable. Several putativeenvironmental agents, such as mercuryand vaccines, have been intenselystUdied, but to date no environmental

agent has been identified to explain therise in incidence. Genetic and genomicstUdies have also failed to explain thense.

Prevalence stUdies assessing timetrends are subject to inaccuracies thatresult from changes in diagnosticcriteria, inability to validate diagnosesand increased awareness of the disorder

across time, all of which have occurred

in autism. I? 19.20 The broadening ofthe diagnostic criteria can be seen inthe revisionsof the DSM since infantileautism was first included in 1980. Theuse of ASD has broadened the criteria

still further. Public and physicianawareness has grown substantially.Validating the AD or ASD is shortcoming in many prevalence stUdies. Asecond stUdy of Minnesota children(MS) illustrates the problems. Thisstudy found a greater increase inprevalence, to 52 per 10,000, than theOC stUdywhich measured incidence.21This difference can be explained inpart by the method of ascertainingsubjects and validation differencesbetween the two stUdies. In the OC

study, investigators reviewed medicaland school records to determine the

diagnosis of each subject using DSMcriteria. The MS stUdy used studentdata reported to the department ofeducation. Such data do not allow

diagnostic validation.The possibility that school data

may be biased towards categorizingstudents as autistic is suggested by thetiming of the initial rise in prevalence(between 1991 and 1994 in moststUdies). Coincident with the initialrise was the inclusion of autism in the

list of disabilities eligible for federallymandated special education services

Table 2. Examples of disorders associated withautistic disorder.

in utero exposure to rubella, CMV, valproate, thalidomide

neonatal or early infant insults (e.g., hypoxia, infection, trauma,hypothyroidism)

brain malformations

chromosomal abnormalities (5%)

Gene mutations (1-2%)15q11-13 (Prader Willi/Angelman region) most frequentfragile X mutationtuberous sclerosisphenylketonuria (if untreated)

in 1991. If this biased diagnosis,stUdies ascertaining cases using schooldata would be expected to identifya rise in prevalence. Support forthis possibility would be evidence ofdiagnosis swapping. Croen20 foundevidence of diagnosis swapping inCalifornia, a state that has reporteda dramatic rise in autism. A periodof rising prevalence in AD saw acorresponding drop in the prevalence ofmental retardation without autism.

The reasons for the disturbing risein the prevalence of AD andASD do notsuggest a meaningful rise in new casesbut a rise due to diagnosis swapping,an increased identification of cases

and a broadening of diagnostic criteriato include cases that were considered

below the diagnostic threshold in thepast. However, the possibility that aportion of the rising prevalence resultsfrom a true increased incidence inautism has not been excluded.

ETIOLOGY

AD is categorized as eitheridiopathic or secondary when the causeis known. Approximately 90% of thecases are idiopathic. A cause of the AScan be identified in approximately 10%.(Table 2) Chromosomal abnormalitiesaccount for 5% of the secondary

AD. Duplication in the Prader Willi/Angelman region (15qll-13) is themost frequent gene mutation. Untreatedphenylketonuria was responsible aconsiderable portion of secondary AD

in the past but is now rare. Other causesof AD include prenatal agents such asinfection and hypoxia. Althoughseveral environmental agents have beenimplicated as causative agents (e.g.,mercury), stUdies have not providedsupporting evidence.

Current conceptUalization of therole of genes in AD is that of impartingsusceptibility but not causing AD.Genetic heterogeneity is the operatinghypothesis in studies with estimatesof the number of involved genesranging from 10-15. Four candidatesusceptibility genes are currently understUdy.22 (Table 3) Linkage stUdieshave identified chromosomes 2, 7, 17,

22 and X, but confirming stUdies havegiven inconsistent results. Numerouscandidate genes have been identifiedincluding most neurotransmitterreceptors and proteins for essentialbrain development (e.g., neuroligin,reelin), but no candidate gene hasbeen consistently found in childrenwith AD. This lack of clarity isnot inconsistent with overlappinggene effects. Genes can be epistatic(several genes influencing a behavior)or pleiotrophic (one gene influencingseveral behaviors).

INHERITANCE

The inheritance of idiopathic ADis complex and non-mendelian. Nosingle inheritance pattern has beenrecognized. Inheritance is thought to bemultifactoral involving the interaction

VOL. 88 NO.5 MAY2(

Table 3. Candidate autism susceptibility genes

AUTS1AUTS2AUTS3AUTS4

chromosome locus7q11 (William syndrome region)3q25

13q1415q11 (Prader-Willi/Angelman syndrome region)

of genes and epigenetic factots. Theevidence for a genetic component inthe neuropathogenesis of AD comesfrom tWin and family studies. Twinstudies have demonstrated a higherconcordance in monozygotic twinsthan dizygotic. 23 Furthermore, somefamily pedigrees show an increasedrecurrence risk within families with onechild with AD.24

The recurrence risk to siblings of achild with secondary AD is the risk ofthe disorder causing AD (e.g., tuberoussclerosis). Determining the empiricrisk of idiopathic AD depends on theincidence of AD which is debated (see

Prevalence). Barbaresi reported anincidence of 4.5/1 0,000.17 Using thisincidence, the empiric risk is less than0.1 %. Some have recommended usingprevalence which would raise the risk to0.5-0.7%. The risk in ASD is debated

but is presumed to be greater thanAD. The risk increases considerablyin families with one child with AD

(4%) and still more when there are tWoor more children with AD (35%).24Because of the association betWeen AD

and language, social and psychiatricproblems, families with one child withAD are given an additional risk of 4-6%for one of these problems.

f6

NEUROPATHOGENESIS

Kanner was the first to speculatethat AD had a biological cause. Hesupported his speculation by observingthat severalchildren had largeheads. Hisobservation of macrocephaly has beenconfirmed by anthropomorphic andneuroimaging studies. Macrocephaly ispresent in 20% of the children with AD,appears late in the first year and resolvesby five years old in most children. Thehigh incidence of seizures and mentalretardation provides further evidencefor a neuropathogenesis.

StUdies to elucidate abnormal

brain regions in AD have produced

inconsistent results while implicatingnumerous cerebral, cerebellar and

brainstem regions. Failure toreplicate neuroanatomical studies,in part because of the unavailabilityof postmortem brain tissue, hashampered an understanding of howbrain development is disrupted. Theetiologic heterogeneity of AD may alsoexplain inconsistencies betWeenstudies.Functional magnetic resonanceimaging (fMRI) offered investigatorsone method of circumventing the

paucity of brain tissue although it isstill limited by etiologic heterogeneity.Despite this, it has provided a meansof correlating brain dysfunction withbrain structures.

Since a hallmark of AD is impairedsocial components of language andcognition (discussed by Dr. Sheinkopfin this issue), the "social brain" regionhas received particular attention. 25Social cognition can be thought of asthe ability to recognize, manipulateand behaviorally respond to socialinformation whether in the form

of language, another's behaviors orexpressions. An important pathwaymediating social cognition involvesthe amygdala, superior temporalsulcus and fusiform gyrus of theorbitofrontal cortex. fMRI studies havedemonstrated the abnormal activation

of these structUres in high functioningautistic individuals.

Of particular interest is the serotoninsystem: 30% of children with AD haveelevated platelet serotonin. Whether thiscontributes to the neuropathogenesis ofAD or is an incidental finding will

require further stUdy.Studies of the autistic brain have

produced heterogeneous collectionsof findings most likely the result ofits heterogeneous etiologies. Thepicture that appears to be emerging isdisruption of early brain developmentwith subsequent abnormalities in

neuronal morphology and number,synaptic abnormalities and dysfunctionalpathways as a consequence.

REFERENCES1. Kanner L. Nervous Child 1943;2:217-50.

2. Bettelheim B. The Empty Fortress: InfantileAutism and the Birth of Self New York: FreePress; 1967.

3. DeMyer MK, Pontius W, et al. ] AutismChildhood Schizoph 1972;2:49-66.4. Diagnostic and Statistical Manual of MentalDisorders, 3rd edn. Washington DC: AmericanPsychiatric Association; 1980.5. Diagnostic and Statistical Manual of MentalDisorders, 4th edn. Washington DC: AmericanPsychiatric Association; 1994.6. Constantino ]N, Todd RD. Arch Gen

Psychiatry2003;60:524-30.7. Piven], Palmer R et at. Am] Psychiatry1997;154:185-90.

8. Spiker 0, Lotspeich L], et at.. Am] Med Gent2002; 11 il4: 129-36.

9. Schultz R, Gauthier I, et at. Arch GenPsychiatry 2000;57:332-43.10. Rivito ER, Freeman B], et al.. Am]Psychiatry 1989;146:194-9.11. De Giacomo A, Fombonne E.. Eur ChildAdolesc Psychiatry 1998;7:131-6.

12. Bertrand], Mars A, et at. Pediatrics2001;108:1155-61.

13. Bryson SE, Clark BS, Smith 1M. ChildPscyhol Psychiatry 1988;29:433-445.14. Bailey A, Palferman S, et al. ] Autism DevDisord 1998;28:369-92.

15. Piven ]. Palmer P. Am ] Psychiatry1999;156:557-63.

16. Yeargin-Allsopp M, et aI.]AM4 2003;289:49-55.17. Barbaresi W, Katusic S, et al. Arch PediatrAdolesc Med2005;159:37-44.18. Issue Brief. Rhode Island Kids Count.December 2003.

19. Wing L, Potter D. Ment Retard Dev DisabilRes Rev 2002;8: 151-6 I.

20. Ctoen LA, Grether ]K, et al. ] Autis DevDisord2002;32:207-15.21. Gurney]G, Fritz MS, et aI. Arch PediatrAdolesc Med2003;157:622-7.22. Online Mendelian Inheritance in Man.

http://www.ncbi.nlm.nih.gov/ entrez/ query.fcgi?db=OMIM (enter aurism)23 Bailey A, Le Couteur A, et al. PsycholMed1995;25:63-78.24. Wassink TH, Brzustowicz L, et aI. MRDDRes Rev 2004;10:272-83.

25. Brothers L. Concepts Neurosci 1990; 1:27-51.

Joseph H Hallett, MD, is Physician-

in-Chief, Department of Pediatrics,Memorial Hospital of Rhode Island, andAssociateProfessor{Clinical}of Pediatrics,Brown Medical School.

VirenD'Sa, MD, isa DevelopmentalBehavioral Pediatric Fellow,

Neurodevelopmental Center, Department

of Pediatrics, Memorial Hospital

Correspondence:Joseph J. Hallett, MD

MEDICINE AND HEAL.TH I RHODE ISL.AND