International Child Development Resource Center

0

• IMFAR CONFERENCE ABSTRACT – SLIDE SESSION 1

S1.1.1
A COMPARISON OF NARRATIVE DISCOURSE IN CHILDREN WITH AUTISM AND SLI. H. Tager-Flusberg*, K. Condouris, and L. Evancie. Lab of Developmental Cognitive Neuroscience, Boston University School of Medicine, Boston, MA 02118.

Recent studies have shown that there is a subgroup of children with autism whose language impairments parallel those found in a different developmental language disorder SLI. The purpose of the study was to examine these parallels further, to identify aspects of language impairment (LI) shared by these groups and those that are unique to autism. Participants included 21 high-functioning children with autism aged 6-14, and 22 children with a history of LI matched on age, nonverbal IQ and vocabulary. Two narratives were collected from each participant using stories that have been widely used in studies of children: (1) a story narrated by the child from a wordless picture book; (2) a retelling of a different story following the narration provided by the examiner. The stories were taped and later transcribed. In addition, the children were administered a battery of theory of mind tasks. The narratives were coded for a number of different features. These included: length, number of different words used, grammatical errors, use of pronouns, mental state words, and story structure. Findings revealed an interesting pattern of similarities and differences between the groups. Discussion will focus on highlighting those aspects of narrative deficit that are unique to autism and their relationship to theory of mind abilities.

S1.1.2
A REGION IN RIGHT PREFRONTAL CORTEX IS ACTIVATED SELECTIVELY BY SEMANTIC PROCESSING OF WORDS ABOUT THE MIND IN NORMAL BUT NOT AUTISTIC INDIVIDUALS C. F. Chabris, I. Aharon, J. A. Clark, L. McGrath, S. Steele, H. Tager-Flusberg, and G. J. Harris* Department of Radiology, Massachusetts General Hospital and Harvard Medical School, Boston, MA, USA.

The right prefrontal cortex plays a critical role in the cognitive system that represents and reasons about the mental states of others; e.g., damage here impairs performance on tests of theory-of-mind ability more severely than does damage elsewhere in the brain. We compared processing of words about the mind (e.g., think, believe, confusion) and other abstract words (e.g., cause, attempt, liberty) that were matched for frequency, length, and reading level. Groups of normal and high-functioning autistic adult males performed semantic (“Is the word positive or negative?”) and perceptual (“Is the word in upper or lower case letters?”) tasks while viewing blocks of these words (and a set of matched concrete words) during whole-brain FMRI at 1.5T over four runs totaling approximately 18 minutes. As predicted, in the normal group, a cluster in the right inferior and orbital frontal gyri was more active during semantic processing of words about the mind than other abstract words. Critically, the autistic group did not show the same activation difference, and neither group showed it during perceptual processing. These results suggest that (1) different categories of abstract words and concepts are processed in different brain regions, (2) neural systems involved in understanding the minds of others are also involved in understanding words about the mind, and (3) underuse of words about the mind by autistic individuals is related to their impaired theory-of-mind ability by a common neural locus.

S1.1.3
TWO STANDARDIZED TESTS OF PRAGMATICS DISTINGUISH CHILDREN WITH ASDS FROM CONTROLS MATCHED FOR LANGUAGE FUNDAMENTALS. E. C. Young, J. J. Diehl, D. Morris, and S. L. Hyman. Departments of Surgery, Clinical and Social Psychology, and Pediatrics, University of Rochester, Rochester, NY 14642

Pragmatic language disorder (PLD) is a defining characteristic of Autism Spectrum Disorders (ASD), but this language parameter is difficult to diagnose and quantiffy. PLD is typically assessed by observation, interviews, and profiles that do not yield quantitative data. This study was done to determine whether standardized tests would differentiate pragmatic language skills in subjects with ASD when they were compared to controls matched for other language skills as assessed by the CELF-3. Fourteen pairs matched for age, verbal IQ, and language participated. They were given the Test of Pragmatic Language (TOPL) and the Strong Narrative Assessment Procedure (SNAP). SNAP measures language skills, including pragmatic functions, in a story retelling format. When scores were compared, the children with ASDs performed significantly worse than controls on the TOPL. Children with an ASD diagnosis also exhibited a significant deficit when answering inferential questions, but not factual questions, on the SNAP. No significant differences were found on syntax, cohesion or story grammar on the SNAP, but there was a trend for ASD subjects to have lower scores on descriptions of the actors’ reactions and problem-solving strategies. The results indicate that deficits in pragmatics and inference are important features of the ASD phenotype, even when subjects and controls are matched for verbal IQ and language skills.

S1.1.4
EMPATHIC ACCURACY IN ADULTS WITH A PERVASIVE DEVELOPMENTAL DISORDER DURING AN UNSTRUCTURED CONVERSATION WITH OTHER PERSONS K. Ponnet, A. Buysse, H. Roeyers, K. De Corte Ghent University, Research Group Developmental Disorders, Ghent, Belgium, B-9000

In the present study, we investigated the degree to which adults with a pervasive developmental disorder (PDD) are able to take the perspective of others, while having an initial conversation with a normally developing person. For this aim, the naturalistic empathic accuracy task was used. Empathic accuracy is the degree to which someone is able to accurately infer the specific content of another person’s thoughts and feelings (Ickes, Stinson, Bissonnette, & Garcia, 1990). As described by Ickes (1997), the most unequivocal way to measure empathic accuracy is by rating the similarity between the content of the target’s actual thought or feeling and the content of the perceiver’s inference. In this study, eleven adults with PDD participated and were videotaped with a concealed camera while having an initial conversation with a normally developing stranger. Using the empathic accuracy design enables the participants to report not only their own thoughts and feelings, but also their inferences about their partner’s thoughts and feelings. Therefore we were able to assess the degree to which persons with and without PDD are accurate in inferring each other’s thoughts and feelings. We further explored the nature of the dyad members’ original thoughts and feelings and assessed some behavioural characteristics (e.g. verbalization and gazing) of both interacting persons. Our findings suggest that there are no differences in empathic accuracy among the participants with or without PDD. These findings suggest that, under some circumstances, adults with PDD are able to take the perspective of others during an initial conversation.

S1.1.5
FEATURES OF LANGUAGE AND COMMUNICATION IN AUTISM: FREQUENCY OF ABNORMALITIES ON ADI-R ITEMS IN THE AGRE SAMPLE. P.E. Iversen, S.J. Spence, J.Y. Yuan, M. Alarcon, A.V. Fedele, AGRE consortium, R.M. Cantor, UCLA Departments of Human Genetics and Psychiatry, Cure Autism Now and AGRE, Los Angeles, CA, 90095.

While autism is defined by three core domains, those affected can display extremely varied behavioral profiles. We are investigating which specific behaviors show an abnormality in the large majority of this heterogeneous population, suggesting a type of common or “universal” feature. One core domain is a deficit in language and communication. The ADI-R (Lord et al, 1994) queries the caretakers of autistic individuals on 31 items specifically related to features of communication. 29 items are rated on an ordinal scale that ranges from relatively normal to severely affected. We set the criterion for a “universal” feature as appearing abnormal (e.g. a score of 1 or more) in greater than 90% of the population. These 29 items were tested in 569 individuals, from the Autism Genetic Resource Exchange (AGRE), who met ADI-R criteria for autism. When adjusting the sample for scores indicating the question was not applicable or not asked, or the answer was not known, nine of these items manifested an abnormality in greater than 90% of the sample. These include items related to the ability to use language to communicate, complexity of non-echoed utterances, social vocalization, reciprocal conversation, expression of interest in others, imitation of actions, gestures, and attention to voice. Only about half of these items are included in the diagnostic algorithm. The two features (reciprocal conversation and expressing interest in others) that exhibited the strongest presence (>98%) were limited to the sample of individuals with higher speech levels. The combined score on these “universal” items provides a good candidate phenotype for studies of familiality and quantitative linkage analysis.

S1.1.6
LANGUAGE AND COGNITIVE FLEXIBILITY IN PEOPLE WITH AUTISM AND DOWN SYNDROME Sophie Jacques, Oriane Landry, Michelle Sutton, Natalie Russo, & Jacob A. Burack, Dept. of Educational Psych., McGill Univ., 3700 McTavish St., Montreal, QC, H3A 1Y2, CANADA

Vygotsky (1986) argued that the emergence of language in the preschool years leads to a revolution in human thought. It is well documented that people with autism are significantly impaired on language relative to their nonverbal abilities (Tager-Flusberg, 2001). Given their uneven developmental profile, they provide an important test for determining which cognitive abilities may depend on language. It has been amply demonstrated that people with autism also have difficulties with tasks that require cognitive flexibility (Baron-Cohen et al. 1985; Hughes & Russell, 1993; Ozonoff et al. 1991), the ability to take simultaneously multiple conflicting social or nonsocial perspectives on a single object or event. Jacques and her colleagues (2001, Jacques & Zelazo, 2001) developed the Flexible Item Selection Task (FIST), a task that minimizes abilities other than cognitive flexibility (existing tasks require other abilities in addition to cognitive flexibility), and they showed that performance on FIST not only relates to language abilities of typically developing preschoolers, but that performance can also be influenced by experimental language manipulations. To determine whether cognitive flexibility varies with language abilities in atypical development, we gave the FIST to 14 people with autism (CA = 145 mos; VMA = 47 mos; NVMA = 72 mos) and 9 people with Down syndrome (DS; CA = 175 mos; VMA = 38 mos; NVMA = 51 mos). For both groups, performance on FIST correlated with VMA (autism, r = 0.53, p < .05; DS, r = 0.84, p < .01), and VMA was the single best predictor of FIST performance (better than CA, NVMA, VIQ, or NVIQ), as predicted. S1.3.1DENSE SNP MAPPING AND LINKAGE ANALYSIS OF PHENOTYPIC SUBSETS INCREASES EVIDENCE FOR AN AUTISM SUSCEPTIBILITY GENE IN 15q11-q13. EL Nurmi, LM Olson, M Dowd, T Amin, MM Jacobs, JL McCauley, AY Lam, SE Folstein, JL Haines and JS Sutcliffe*. Program in Human Genetics, Vanderbilt University, Nashville, TN 37232 and Dept of Psychiatry, Tufts Univ, Boston, MA 02111.

Chromosome 15q11-q13 is a candidate region for autism susceptibility based on chromosomal duplications in autism and linkage disequilibrium (LD) at 15q markers, however significant linkage has been inconsistently reported. To increase power for detecting genetic effects, we analyzed linkage in phenotypic subsets in the CLSA families based on a factor analysis of the ADI. We identified significantly increased evidence for linkage in a subset, which has splinter skills (HLOD 2.6 at 9.8 cM). To localize genetic susceptibility, we developed and analyzed a dense SNP map of this region. Initially we focused on the imprinted domain containing the maternally-expressed genes UBE3A and ATP10C and putative imprinting control regions. With a marker coverage of 1/10 kb in coding regions and 1/15 kb in large 5¥ introns, this map was employed to dissect LD in autism families. Four SNPs within ATP10C demonstrated evidence for preferential allelic transmission to affected offspring. All four SNPs lie within islands of human-mouse sequence homology. While these SNPs may be functionally significant, it is more likely that they represent the detection of an ancestral haplotype containing a functional susceptibility allele. The region was further explored for recombination hot spots and haplotype blocks to evaluate haplotype transmission in our families. Taken together, these findings provide increased evidence for an autism susceptibility locus in 15q11-q13.

S1.3.2
LINKAGE OF AUTISTIC DISORDER TO CHROMOSOME 15Q11-Q13 USING PHENOTYPIC SUBTYPES Y. Dementieva 1, Y. Shao1, M.L. Cuccaro1, E.R. Hauser1, K.L. Raiford1, C.M. Wolpert1, S.L. Donnelly1, L.A. Elston1, S.A. Ravan2, R.K. Abramson2, H.H. Wright2, G.R. DeLong1, J.R. Gilbert1, M.A. Pericak-Vance1. 1) Center for Human Genetics, Duke University Medical Center, Durham, NC 27710; 2) W.S. Hall Psychiatric Institute, University of South Carolina, Columbia, SC 29202.

S1.4.1
SCREENING FOR AUTISM SPECTRUM DISORDERS: THE ACCURATE USE OF CLINICAL CLUES. J.M. Gillis and R.G. Romanczyk*. Institute for Child Development, SUNY at Binghamton, Binghamton, NY, 13902.

The reported prevalence of Autism Spectrum Disorders (ASD) has increased substantially. Estimates have ranged from 3-5 per 10,000 to over 20 per 10,000 (Fombonne, 1999). Even with the variation in prevalence, there is agreement that early identification is critical. Relatedly, evidence for the effectiveness of intensive early intervention is strongly supported (National Research Council, 2001). Accurate use of screening instruments is essential in the early identification process. The present study investigated the clinical clues endorsed by the New York State Department of Health (NYSDOH, 1999), intended for use by a range of professional and paraprofessional service providers, not necessarily familiar with ASD. Participants viewed videos of children with ASD and identified specific clinical clues. Four experimental groups crossed the factor of information (developmental milestones or information about early intervention) and the factor of level of experience with children with ASD (at least six months or none). Results indicated experience wiith children with ASD improve accuracy compared to participants with no experience. Information regarding developmental norms did not improve the accurate use of the clinical clues. Thus, it appears that when training paraprofessionals or professionals in the use of screening instruments for children with ASD, relevant information alone is insufficient to produce accuracy. Rather experience with children with ASD is an essential element. However, significant rates of false positives among the most accurate group indicate that practical training must address the complex issue of accuracy even for “simple” screening instruments.

S1.4.2
THE INTERRELATIONSHIP BETWEEN ADOS-G, ADI-R AND DSM-IV-TR CLASSIFICATION IN CHILDREN AND ADOLESCENTS WITH MENTAL RETARDATION A. de Bildt, S. Sytema, C. Ketelaars, D. Kraijer, E. Mulder, F. Volkmar, R. Minderaa. Child and Adolescent Psychiatry, University of Groningen, 9713 GZ, Groningen, the Netherlands

Classification of PDD is known to be complicated in children and adolescents with MR, due to behavioral overlap between PDD and MR, and low mental ages in MR that account for impaired social and communicative behavior. This study aimed to investigate the value of the ADI-R and ADOS-G in the diagnostic process of PDD in children and adolescents with MR. The interrelationship between the ADI-R, ADOS-G and clinical DSM-IV-TR classification was studied in children and adolescents with MR (N=184, all levels of MR, age 5-20). The level of agreement on classification, sensitivity, specificity and construct-related validity of the ADI-R and ADOS-G were studied compared to the clinical DSM-IV-TR classification. The results indicate a low level of agreement between ADI-R and ADOS in classifying PDD. The validity and reliability of the ADI-R and ADOS were satisfactory. The level of agreement between the instruments will be discussed with respect to source of information and concept under study. The reliability and the validity will be discussed with respect to clinical use of the instruments in the diagnostic process of PDD in children and adolescents with PDD.

S1.4.3
SENSORY-MOTOR AND SOCIAL-COMMUNICATIVE PREDICTORS OF AUTISM IN INFANCY. G.T. Baranek, L. Watson, E. Crais, J. C. Neitzel, and S.Barry. Department of Allied Health Sciences, University of North Carolina at Chapel Hill, School of Medicine, Chapel Hill, NC 27599.

There is a need to define the nature of the earliest symptoms of autism, and the trajectories of symptoms to develop criteria that permit earlier, reliable diagnosis. This paper presents findings from a new pilot study using retrospective video methodologies to analyze sensory-motor and social-communicative behaviors at two different age points in children with autism — early infancy (9-12 months of age) and later infancy (15-18 months). The aims are to: Clarify the role of sensory-motor behaviors relative to social-communicative behaviors as early predictors of autism. Examine the variability and developmental changes in symptomatology of autism from early infancy (9-12 months) to later infancy (15-18 months). Establish the relationship between early sensory-motor symptoms and later social-communicative symptoms. Preliminary results will include data on 3 groups of subjects (autism, DD, typical). Currently, 16 subjects’ data are available for analyses, however we expect to incorporate 10 additional subjects into these analyses prior to the presentation. Home videos were edited to obtain a 10-minute cross-section of situations for each subject at both time points. All footage was coded for frequencies of behaviors in several mutually exclusive behavioral categories (looking, affect, motor/object stereotypies, response to name, social touch, visual orientation, and gestures). Data will be analyzed with descriptive statistics and discriminant or logistic functions to determine patterns of symptoms that best predict a diagnosis of autism at each of the two separate time points (early and late infancy). Correlational analyses will be used to determine the association of early sensory-motor symptoms to later social-communicative symptoms. Methodological limitations and implications for future research will be explicated.

S1.4.4
EARLY DETECTION OF AUTISM: EVIDENCE FROM INFANT SIBLINGS OF CHILDREN WITH AUTISM Landa, R.J. Kennedy Krieger Institute, 3901 Greenspring Avenue, Baltimore, MD 21211.

Early detection of autism is needed to facilitate early intervention. There is evidence that the brain abnormality in autism has a prenatal onset. Since autism is heritable, and the recurrence risk for autism has been estimated to be 4 to 10%, we studied infant siblings of children with autism. The objectives of this study were to identify early behavioral markers for autism spectrum disorder and to determine developmental trajectory of autism in toddlers between 14 and 24 months of age. We hypothesized that the early expression of autism would manifest as relative impairments in joint attention, shared positive affect and specific communication features. We examined 45 siblings of autistic probands (autism sibs) and 29 controls. Social cognition, play, and language were assessed. Lack of comprehension of pointing gestures at 14 months was predictive of ASD at 24 months (p=0.04). Language comprehension and production, play, joint attention, and shared positive affect (SPA) were major features in the early manifestation of ASD (14 months of age), consistent with the literature (e.g., Baranek, 1999; Baron-Cohen et al., 1996; Osterling & Dawson, 1994; Stone et al., 1999) and support our hypothesis that early impairment in interpersonal synchrony is predictive of ASD. Developmental trajectory of children with autism from 14 to 24 months showed static or regressive patterns compared to those with non-autism language impairment in the domains of social affective and social cognitive domains. This is the first longitudinal study of autism in siblings of children with autism beginning at 14 months of age. Our data indicate that, in many cases, autism can be diagnosed by 14 months of age.

S1.4.5
PREDICTING SOCIAL-COMMUNICATION IMPAIRMENTS IN YOUNG SIBLINGS OF CHILDREN WITH AUTISM. L. Zwaigenbaum, S.E. Bryson, J. Brian, C. McDermott, V. Rombough, W. Roberts, P. Szatmari. Centre for Studies of Children at Risk, McMaster University, Hamilton, Ontario L8N 3Z5 and Autism Research Unit, The Hospital for Sick Children, Toronto, Ontario M5G 1X8.

Background: To better inform early identification efforts, we have initiated a prospective study of high-risk infants, defined by virtue of having a sibling with autism. A sample of 49 infant siblings has been followed, as well as 20 control infants with no family history of autism. Our main goal is to assess whether behavioural markers in the first year of life predict later impairments. Methods: Infants are assessed using the Autism Observational Scale for Infants (AOSI), which includes 18 items (or ‘risk markers’) derived from retrospective reports, home videotape studies and other sources. The test-retest and inter-observer reliability of the AOSI at 12 months are excellent (intraclass correlation 0.66 and 0.92 respectively). We compared risk markers in infant siblings and controls at 12 months, and assessed whether risk markers at 12 months predict clinical impairments at 18 months. Results: 1) 12-month-old siblings demonstrate a higher number of risk markers (positive AOSI items) than controls. 2) Total risk marker count at 12 months is highly predictive of later impairments, accounting for 39% of variance in ADOS (social + communication) algorithm scores in siblings at 18 months. 3) Five of 40 siblings exceeded ADOS algorithm threshold for autism at 18 months, including 4 of 6 siblings with 6 or more markers at 12 months, versus 1 of 34 siblings with fewer markers and no controls. Conclusion: Behavioural risk markers measured by the AOSI at 12 months of age are predictive of later social-communication impairments in high-risk infants.

S1.4.6
DISENGAGEMENT OF VISUAL ATTENTION IN 6-MONTH-OLD SIBLINGS OF CHILDREN WITH AUTISM. T. Rogers, S.E. Bryson* and L. Zwaigenbaum*. Autism Research Unit, The Hospital for Sick Children, Toronto, Ontario, CANADA, M3H 3B1.

Background: Retrospective reports suggest that signs of autism may be present during the first year of life. However, no specific marker has been evaluated prospectively. One potential candidate is the disengage function of visual attention: children with autism have been known to shown to be impaired in disengaging from one of two competing stimuli, an ability that is well established in typically developing infants by 4 months of age. We hypothesize that impaired disengagement in infancy may be a risk marker for autism. As a preliminary test of this hypothesis, we provide data on visual disengagement in high-risk 6-month-olds (siblings of children with autism; N = 20) compared to low-risk controls (negative family history; N = 20). Method: On each trial, infants were first visually engaged on a central stimulus before a competing peripheral stimulus was displayed. Latency (in msecs) to begin an eye movement to the peripheral stimulus served as the dependent measure. Results: The sibling group took significantly longer than the comparison group to disengage and shift gaze to the peripheral stimulus (M = 737, SD = 404 msecs; control: 531, SD = 218 msecs; F = 4.03, p < .05). 7 of 20 sibs failed to disengage on at least 1 trial, compared to only 1 of 20 controls (F = 5.6, p = .02). Continued follow-up of this high-risk group will help determine whether an early disengage impairment is predictive of a later diagnosis of autism.

SLIDE SESSION 1.2: NEUROPSYCHOLOGY & NEUROIMAGING

S1.2.1
FUNCTIONAL ANATOMY OF COMPENSATORY PROCESSING IN AUTISTIC ATTENTION: COMPLEMENTARY ROLES OF SELECTION AND SUPPRESSION. MK Belmonte and DA Yurgelun-Todd*. Cognitive Neuroimaging Laboratory, McLean Hospital, Belmont, MA 02478.

Despite a deficit in rapid re-allocation of attention, and absence of normal physiological indices of attentional selection, people with autism display intact and even superior performance on many attentional tasks. When attentional selection is impaired developmentally, compensatory cognitive strategies may be brought into play in order to achieve functional behavioural performance. On the basis of an fMRI study of visual spatial attention, we previously reported that in autism, generalised arousal substitutes for early selective attention. In an extension of this work, we have localised in intraparietal sulcus a process of suppression of irrelevant stimuli which may compensate for impaired selective processing. Subjects viewed coloured squares flashed at 9 Hz in the left and right upper quadrants. A red square in the attended location cued not only an overt behavioural response but also a covert shift of attention to the opposite hemifield. In echo-planar images acquired at 1.5T, effects of leftward and rightward directed attention were compared using a permutation test within regions of interest defined a priori on the basis of individual functional maps. In comparison with a group of normal subjects matched for age and sex, a group of 6 subjects with autism showed less attention-related activity in ventral occipital cortex contralateral to the attended hemifield (p < 0.03), and greater attention-related activity in intraparietal sulcus contralateral to the suppressed hemifield (p < 0.04). In addition to these findings, we will report preliminary measurements on frontal regions.

S1.2.2
ORBITOFRONTAL DYSFUNCTION IN AUTISM. K.A. Loveland*, J. Bachevalier, D.A. Pearson, D. Lane, E. Nagy, J. B. Shaw, S. Reddoch, K. Dodds. Ctr. Human Dev. Research, Dept. Psychiatry & Beh. Sci., Univ. TX Med. Sch., Houston, TX 77030

Functioning of the orbitofrontal cortex (OFC) was assessed in 3 groups age 7 to 18 (Autism Hi IQ (HFA) n=30; Autism Low IQ (LFA) n=40; Control Hi IQ (HFC) n=17) and a Hi IQ Preschool Control group (Pr; n=10) age 3 to 6. Subjects received two tasks of object discrimination reversal (ODR) graded in difficulty (1-pair, 5-pair versions) with a discrimination phase and six reversals each. ODR tests ability to rapidly adjust responses to the changing reward value of stimuli, a function subserved by OFC. Group differences in mean errors were compared with repeated measures ANOVA (Group (4) x Task (2) x Reversal (7)), with the effect of reversals assessed by linear component of trend. The interaction of Task x Reversal x Group was significant (F3,93=5.237, p=0.002), as were Reversal x Group (F3,93=2.694, p=0.05) and Task x Group (F1,93=7.40, p=0.001). Follow-up comparisons (SNK) showed Pr and LFA groups made significantly more errors than HFA and HFC on both tasks (p S1.2.3
PREFRONTAL ACTIVATION DURING PERCEPTION OF NEGATIVE EMOTION IN CHILDREN AND ADOLESCENTS. D Yurgelun-Todd*, L Femia, A Young, Brain Imaging Center, McLean Hospital/Harvard Medical School, Belmont, MA 02478.

Autism is a behaviorally defined disorder having significant impairments in social interactions and facial processing. Recent findings suggest these impairments are associated with the inability to abstract and generalize in the social domain. The extent to which these functional deficits are related to developmental brain changes is not known. Most affect perception studies of adults report greater activation in right prefrontal cortex and the left amygdala, whereas the few studies of emotional perception in healthy children and adolescents suggest that lateralization may shift dynamically during development. We examined processing of facial affect in children and adolescents using BOLD fMRI. Fourteen healthy participants (age 8-15) were presented with two stimulus conditions viewing happy and fearful faces. Twenty coronal images were acquired on a GE LX 1.5 Tesla scanner using echo planar imaging. A 64 x 64 image matrix was used with a 3mm x 3mm in-plane resolution and a 7 mm slice thickness. Cerebral activation was analyzed in SPM99 with a whole-brain voxel-wise correction for multiple comparisons (p < .05). Subtraction of activation during the happy condition from the fear condition yielded significant (p = .003) activation in the left and right prefrontal cortex. However, subtracting fear from the happy condition yielded no significant activation. Our findings suggest that in healthy children and adolescents fear perception may be processed in bilateral prefrontal cortex in contrast to the right prefrontal activation seen in adults. These results imply a transition of emotional function from the dominant to non-dominant hemisphere during development and may provide insight into behavioral deficits in autism.

S1.2.4
NEUROIMAGING RELATIONSHIPS TO BEHAVIORAL PERFORMANCE AND CLINICAL COURSE IN YOUNG CHILDREN WITH ASD S Dager*, J Munson, S Friedman, S Webb, D Shaw, B Sparks, A Artru, R Abbott, G Dawson University of Washington, Seattle WA 98105

Autism Spectrum Disorder (ASD) is a heterogeneous disorder with varying prognosis and behavioral deficits. To explore the functional significance of neuroimaging findings, including previously reported structural and neurochemical abnormalities associated with ASD, relationships to behavioral characteristics and clinical course were examined. Children aged 3- to 4-years with ASD (N=45) underwent MRI for volumetric analyses of brain anatomy and proton echo-planar spectroscopic imaging (PEPSI) for tissue-based regional brain chemical quantification. Neuropsychological test batteries that tapped nonverbal IQ, specific brain regions, such as medial temporal lobe and prefrontal cortex,and autism-related behaviors were administered to 3- to 4-year-children with ASD and reevaluated as part of a 3-year longitudinal follow-up. Overall, cerebral and cerebellar volumes were not correlated with nonverbal IQ. Several neuropsychiatric measures that reflected severity of autism symptoms, including joint attention, response to emotional cues and imitation ability, were inversely correlated with amygdalar volume at 3 to 4 years of age. In contrast, hippocampal volumes were positively correlated with performance for tasks tapping prefrontal functions. These neuroanatomical and neuropsychological interrelationships, as well as relationships to neurochemical findings, are currently being examined in a multivariate context and in relationship to longitudinal course to better ascertain the functional and prognostic significance of neuroimaging findings in ASD.

S1.2.5
AFFECTIVE STARTLE MODIFICATION IN AUTISM J. L. Wilbarger, D. N. McIntosh*, P. Winkielman, A. Riechman-Decker, & A. Curiel. Emotion and Cognition Lab, Department of Psychology, University of Denver, Denver, CO 80208.

Individuals with autism appear to have deficits in core affective processes that interfere with their ability to respond automatically to affective information. The speed and intensity of the startle response is modified by a person’s underlying affective state; thus, examining affective startle modification in autism provides a way to study potential core affective deficits. Affective modification of startle responses in 8 high functioning adults with autistic spectrum disorders is compared to responses in 7 typically developing individuals. Startle magnitude was measured by recording electromyographic (EMG) activity over the orbicularis oculi during the presentation of 95db startle probes while participants viewed pictures with affective content. Typical adults demonstrated the expected pattern of potentiated startle magnitudes during exposure to negative affective images and decreased intensity during positive affective images, as compared to startle during neutral images. In contrast, individuals with autistic spectrum disorders demonstrated potentiated startle responses during exposure to both positive and negative images. Results can be interpreted as poor automatic discrimination of the affective content of the stimuli. These results might be explained by the alerting value of the stimuli, which is higher for both the positive and the negative stimuli than the neutral stimuli. This interpretation can be supported by analysis measures of general arousal such as electrodermal activity, and additional measures of affective responding such as picture ratings and facial EMG over the corrugator and zygomaticus.

S1.3.1
DENSE SNP MAPPING AND LINKAGE ANALYSIS OF PHENOTYPIC SUBSETS INCREASES EVIDENCE FOR AN AUTISM SUSCEPTIBILITY GENE IN 15q11-q13. EL Nurmi, LM Olson, M Dowd, T Amin, MM Jacobs, JL McCauley, AY Lam, SE Folstein, JL Haines and JS Sutcliffe*. Program in Human Genetics, Vanderbilt University, Nashville, TN 37232 and Dept of Psychiatry, Tufts Univ, Boston, MA 02111.

Chromosome 15q11-q13 is a candidate region for autism susceptibility based on chromosomal duplications in autism and linkage disequilibrium (LD) at 15q markers, however significant linkage has been inconsistently reported. To increase power for detecting genetic effects, we analyzed linkage in phenotypic subsets in the CLSA families based on a factor analysis of the ADI. We identified significantly increased evidence for linkage in a subset, which has splinter skills (HLOD 2.6 at 9.8 cM). To localize genetic susceptibility, we developed and analyzed a dense SNP map of this region. Initially we focused on the imprinted domain containing the maternally-expressed genes UBE3A and ATP10C and putative imprinting control regions. With a marker coverage of 1/10 kb in coding regions and 1/15 kb in large 5¥ introns, this map was employed to dissect LD in autism families. Four SNPs within ATP10C demonstrated evidence for preferential allelic transmission to affected offspring. All four SNPs lie within islands of human-mouse sequence homology. While these SNPs may be functionally significant, it is more likely that they represent the detection of an ancestral haplotype containing a functional susceptibility allele. The region was further explored for recombination hot spots and haplotype blocks to evaluate haplotype transmission in our families. Taken together, these findings provide increased evidence for an autism susceptibility locus in 15q11-q13.

S1.3.2
LINKAGE OF AUTISTIC DISORDER TO CHROMOSOME 15Q11-Q13 USING PHENOTYPIC SUBTYPES Y. Dementieva 1, Y. Shao1, M.L. Cuccaro1, E.R. Hauser1, K.L. Raiford1, C.M. Wolpert1, S.L. Donnelly1, L.A. Elston1, S.A. Ravan2, R.K. Abramson2, H.H. Wright2, G.R. DeLong1, J.R. Gilbert1, M.A. Pericak-Vance1. 1) Center for Human Genetics, Duke University Medical Center, Durham, NC 27710; 2) W.S. Hall Psychiatric Institute, University of South Carolina, Columbia, SC 29202.

Autistic disorder (AutD) is a complex genetic disease. It is postulated that several genes contribute to the underlying genetic risk of developing AutD. However, both etiologic and genetic heterogeneity confound the single gene disorders such as fragile X, maternal duplications of chromosome 15q11-q13, and unknown etiology. The fact that maternal but not paternal duplications of 15q11-q13 cause autism suggests that the known genomic imprinting in the region is relevant to the etiology in these cases. The data are compatible with the hypothesis that over-expression of the Angelman gene (UBE3A) may be the molecular basis of autism in duplication cases (i.e., the Angelman gene is also an autism gene). We propose that various genetic and epigenetic defects in 15q11-q13 might account for a substantial fraction of autism cases. To assess the frequency of 15q11-q13 defects in autism, we performed interphase FISH on probands from 56 multiplex families obtained from AGRE. We detected duplication of 15q11-q13 in two affected sibs and their normal mother. To assess allele sharing for 15q11-q13, we are genotyping the families for 6 microsatellites. We are examining autism brain tissue for evidence of genetic or epigenetic abnormalities of 15q11-q13. If such defects were epigenetic and brain-specific, they would not be detectable using DNA from blood or cultured cells. Thus we are testing the hypothesis that a substantial fraction of autism might be caused by genetic and epigenetic defects of 15q11-q13 that would not be detected by usual tests.

S1.3.5
LANGUAGE QTL ON 7q: AN UPDATE. M. AlarcÛn, D. H. Geschwind, the A.G.R.E. Consortium, and R. M. Cantor. Departments of Neurology and Human Genetics, University of California, Los Angeles CA 90095.

Autism is a neurodevelopmental disorder with an early childhood onset and is characterized by deficits in language and social skills, and repetitive behaviors. We reported evidence of linkage for ‘age at first word’ (item 12 from the Autism Diagnostic Interview-Revised) on chromosome 7q in a sample of 152 families from the Autism Genetic Resource Exchange (AGRE; AlarcÛn et al., 2002). A nonparametric multipoint analysis with the Genehunter program resulted in a linkage peak that localized the QTL (quantitative trait locus) to a 24 cM region between markers D7S1824 and D7S3058. Other items of the ADI-R, ëage at first phrase’ (item 13) and a composite measure of ërepetitive and stereotyped behavior’ (item DD total) were not linked to this region. Additional analyses with 28 flanking markers narrowed the region to 10 cM between D7S1824 and D7S2426. These findings suggested that the putative susceptibility gene on chromosome 7 may be a QTL specific for the language deficits associated with autism. Here, we report the results of adding 114 families and 9 markers to these analyses. Linkage to ‘age at first word’ is supported: the peak for the putative gene contributing to ‘age at first word’ has been narrowed and covers a 5 cM region between D7S676 and D7S2511 (Z = 2.9, p < 0.002). Thus, these results provide additional evidence there is a QTL for ‘age at first word’ in autism on chromosome 7q. The AGRE sample continues to expand as additional families are recruited and characterized. Our most recent analyses of this updated sample will be presented.

S1.3.6
CHROMOSOME 22 ABNORMALITIES DETECTED IN AUTISM PATIENTS: WHEN TO SEND CHROMOSOMES. L.M. Lelis, S.J. Spence, M.A. Fox, B.F. Crandall UCLA School of Medicine, Department of Pediatrics and Psychiatry, Division of Medical Genetics and Pediatric Neurology, Los Angeles, CA 90095

The overall rate of chromosomal abnormalities in autistic patients is about 5-9% (Cook, 2001; Fombonne, 1997; Lewis, 1995; Wassink, 2001). Despite this, karyotype analysis has not traditionally been part of the routine work-up in autism. We report a 3 yo female with an autism diagnosis seen for routine neurological evaluation. She had significant language delay, social difficulties, perseverative behaviors, possible seizures, hyperactivity, sleep problems and hypotonia, all consistent with autism. However, she also had motor delays, ataxia, facial dysmorphic features (low set/protruding ears, epicanthal folds, broad nasal bridge) not usually seen in non-idiopathic autism. After genetics consultation, karyotype analysis, Angelman methylation DNA analysis, and the FISH probe for a distal deletion on chromosome 22 (not the DiGeorge/VCF probe) were sent. Standard karyotype was normal but FISH revealed a distal subtelomeric deletion at 22q13.3. Review of the literature reveals ~ 20 cases reported with this deletion syndrome. The most recent reports describe 4 patients specifically with autistic features (Prasad, 2000; Goizet, 2000). Another describes 2 patients with language delay, hypotonia and some autistic features (poor eye contact, unusual sensory issues)(Precht, 1998). The most consistent features include hypotonia, developmental delay, (gross/fine motor and language), minor facial dysmorphology, sleep difficulty, feeding problems, behavior problems, and normal to accelerated growth (Cassidy, 1999; Narahara, 1992; Nesslinger, 1994; Romain, 1990; Schroder, 1998). We conclude that the subtelomeric chromosome 22q13.3 deletion syndrome is important to screen for in certain autism patients especially those with developmental delay (speech and gross motor), significant hypotonia, minor facial dysmorphic features, and normal or advanced growth.

SLIDE SESSION 1.4: EARLY RECOGNITION & SCREENING

S1.4.1
SCREENING FOR AUTISM SPECTRUM DISORDERS: THE ACCURATE USE OF CLINICAL CLUES. J.M. Gillis and R.G. Romanczyk*. Institute for Child Development, SUNY at Binghamton, Binghamton, NY, 13902.

The reported prevalence of Autism Spectrum Disorders (ASD) has increased substantially. Estimates have ranged from 3-5 per 10,000 to over 20 per 10,000 (Fombonne, 1999). Even with the variation in prevalence, there is agreement that early identification is critical. Relatedly, evidence for the effectiveness of intensive early intervention is strongly supported (National Research Council, 2001). Accurate use of screening instruments is essential in the early identification process. The present study investigated the clinical clues endorsed by the New York State Department of Health (NYSDOH, 1999), intended for use by a range of professional and paraprofessional service providers, not necessarily familiar with ASD. Participants viewed videos of children with ASD and identified specific clinical clues. Four experimental groups crossed the factor of information (developmental milestones or information about early intervention) and the factor of level of experience with children with ASD (at least six months or none). Results indicated experience wiith children with ASD improve accuracy compared to participants with no experience. Information regarding developmental norms did not improve the accurate use of the clinical clues. Thus, it appears that when training paraprofessionals or professionals in the use of screening instruments for children with ASD, relevant information alone is insufficient to produce accuracy. Rather experience with children with ASD is an essential element. However, significant rates of false positives among the most accurate group indicate that practical training must address the complex issue of accuracy even for “simple” screening instruments.

S1.4.2
THE INTERRELATIONSHIP BETWEEN ADOS-G, ADI-R AND DSM-IV-TR CLASSIFICATION IN CHILDREN AND ADOLESCENTS WITH MENTAL RETARDATION A. de Bildt, S. Sytema, C. Ketelaars, D. Kraijer, E. Mulder, F. Volkmar, R. Minderaa. Child and Adolescent Psychiatry, University of Groningen, 9713 GZ, Groningen, the Netherlands

Classification of PDD is known to be complicated in children and adolescents with MR, due to behavioral overlap between PDD and MR, and low mental ages in MR that account for impaired social and communicative behavior. This study aimed to investigate the value of the ADI-R and ADOS-G in the diagnostic process of PDD in children and adolescents with MR. The interrelationship between the ADI-R, ADOS-G and clinical DSM-IV-TR classification was studied in children and adolescents with MR (N=184, all levels of MR, age 5-20). The level of agreement on classification, sensitivity, specificity and construct-related validity of the ADI-R and ADOS-G were studied compared to the clinical DSM-IV-TR classification. The results indicate a low level of agreement between ADI-R and ADOS in classifying PDD. The validity and reliability of the ADI-R and ADOS were satisfactory. The level of agreement between the instruments will be discussed with respect to source of information and concept under study. The reliability and the validity will be discussed with respect to clinical use of the instruments in the diagnostic process of PDD in children and adolescents with PDD.

S1.4.3
SENSORY-MOTOR AND SOCIAL-COMMUNICATIVE PREDICTORS OF AUTISM IN INFANCY. G.T. Baranek, L. Watson, E. Crais, J. C. Neitzel, and S.Barry. Department of Allied Health Sciences, University of North Carolina at Chapel Hill, School of Medicine, Chapel Hill, NC 27599.

There is a need to define the nature of the earliest symptoms of autism, and the trajectories of symptoms to develop criteria that permit earlier, reliable diagnosis. This paper presents findings from a new pilot study using retrospective video methodologies to analyze sensory-motor and social-communicative behaviors at two different age points in children with autism — early infancy (9-12 months of age) and later infancy (15-18 months). The aims are to: 1. Clarify the role of sensory-motor behaviors relative to social-communicative behaviors as early predictors of autism. 2. Examine the variability and developmental changes in symptomatology of autism from early infancy (9-12 months) to later infancy (15-18 months). 3. Establish the relationship between early sensory-motor symptoms and later social-communicative symptoms. Preliminary results will include data on 3 groups of subjects (autism, DD, typical). Currently, 16 subjects’ data are available for analyses, however we expect to incorporate 10 additional subjects into these analyses prior to the presentation. Home videos were edited to obtain a 10-minute cross-section of situations for each subject at both time points. All footage was coded for frequencies of behaviors in several mutually exclusive behavioral categories (looking, affect, motor/object stereotypies, response to name, social touch, visual orientation, and gestures). Data will be analyzed with descriptive statistics and discriminant or logistic functions to determine patterns of symptoms that best predict a diagnosis of autism at each of the two separate time points (early and late infancy). Correlational analyses will be used to determine the association of early sensory-motor symptoms to later social-communicative symptoms. Methodological limitations and implications for future research will be explicated.

S1.4.4
EARLY DETECTION OF AUTISM: EVIDENCE FROM INFANT SIBLINGS OF CHILDREN WITH AUTISM Landa, R.J. Kennedy Krieger Institute, 3901 Greenspring Avenue, Baltimore, MD 21211.

Early detection of autism is needed to facilitate early intervention. There is evidence that the brain abnormality in autism has a prenatal onset. Since autism is heritable, and the recurrence risk for autism has been estimated to be 4 to 10%, we studied infant siblings of children with autism. The objectives of this study were to identify early behavioral markers for autism spectrum disorder and to determine developmental trajectory of autism in toddlers between 14 and 24 months of age. We hypothesized that the early expression of autism would manifest as relative impairments in joint attention, shared positive affect and specific communication features. We examined 45 siblings of autistic probands (autism sibs) and 29 controls. Social cognition, play, and language were assessed. Lack of comprehension of pointing gestures at 14 months was predictive of ASD at 24 months (p=0.04). Language comprehension and production, play, joint attention, and shared positive affect (SPA) were major features in the early manifestation of ASD (14 months of age), consistent with the literature (e.g., Baranek, 1999; Baron-Cohen et al., 1996; Osterling & Dawson, 1994; Stone et al., 1999) and support our hypothesis that early impairment in interpersonal synchrony is predictive of ASD. Developmental trajectory of children with autism from 14 to 24 months showed static or regressive patterns compared to those with non-autism language impairment in the domains of social affective and social cognitive domains. This is the first longitudinal study of autism in siblings of children with autism beginning at 14 months of age. Our data indicate that, in many cases, autism can be diagnosed by 14 months of age.

S1.4.5
PREDICTING SOCIAL-COMMUNICATION IMPAIRMENTS IN YOUNG SIBLINGS OF CHILDREN WITH AUTISM. L. Zwaigenbaum, S.E. Bryson, J. Brian, C. McDermott, V. Rombough, W. Roberts, P. Szatmari. Centre for Studies of Children at Risk, McMaster University, Hamilton, Ontario L8N 3Z5 and Autism Research Unit, The Hospital for Sick Children, Toronto, Ontario M5G 1X8.

Background: To better inform early identification efforts, we have initiated a prospective study of high-risk infants, defined by virtue of having a sibling with autism. A sample of 49 infant siblings has been followed, as well as 20 control infants with no family history of autism. Our main goal is to assess whether behavioural markers in the first year of life predict later impairments. Methods: Infants are assessed using the Autism Observational Scale for Infants (AOSI), which includes 18 items (or ‘risk markers’) derived from retrospective reports, home videotape studies and other sources. The test-retest and inter-observer reliability of the AOSI at 12 months are excellent (intraclass correlation 0.66 and 0.92 respectively). We compared risk markers in infant siblings and controls at 12 months, and assessed whether risk markers at 12 months predict clinical impairments at 18 months. Results: 1) 12-month-old siblings demonstrate a higher number of risk markers (positive AOSI items) than controls. 2) Total risk marker count at 12 months is highly predictive of later impairments, accounting for 39% of variance in ADOS (social + communication) algorithm scores in siblings at 18 months. 3) Five of 40 siblings exceeded ADOS algorithm threshold for autism at 18 months, including 4 of 6 siblings with 6 or more markers at 12 months, versus 1 of 34 siblings with fewer markers and no controls. Conclusion: Behavioural risk markers measured by the AOSI at 12 months of age are predictive of later social-communication impairments in high-risk infants.

S1.4.6
DISENGAGEMENT OF VISUAL ATTENTION IN 6-MONTH-OLD SIBLINGS OF CHILDREN WITH AUTISM. T. Rogers, S.E. Bryson* and L. Zwaigenbaum*. Autism Research Unit, The Hospital for Sick Children, Toronto, Ontario, CANADA, M3H 3B1.

Background: Retrospective reports suggest that signs of autism may be present during the first year of life. However, no specific marker has been evaluated prospectively. One potential candidate is the disengage function of visual attention: children with autism have been known to shown to be impaired in disengaging from one of two competing stimuli, an ability that is well established in typically developing infants by 4 months of age. We hypothesize that impaired disengagement in infancy may be a risk marker for autism. As a preliminary test of this hypothesis, we provide data on visual disengagement in high-risk 6-month-olds (siblings of children with autism; N = 20) compared to low-risk controls (negative family history; N = 20). Method: On each trial, infants were first visually engaged on a central stimulus before a competing peripheral stimulus was displayed. Latency (in msecs) to begin an eye movement to the peripheral stimulus served as the dependent measure. Results: The sibling group took significantly longer than the comparison group to disengage and shift gaze to the peripheral stimulus (M = 737, SD = 404 msecs; control: 531, SD = 218 msecs; F = 4.03, p < .05). 7 of 20 sibs failed to disengage on at least 1 trial, compared to only 1 of 20 controls (F = 5.6, p = .02). Continued follow-up of this high-risk group will help determine whether an early disengage impairment is predictive of a later diagnosis of autism.