American Journal of Medical Genetics Part B (Neuropsychiatric Genetics) 144B:1022 –1026 (2007) Autism in African American Families: Clinical-Phenotypic Findings Michael L. Cuccaro,1* Jason Brinkley,2 Ruth K. Abramson,3 Alicia Hall,3 Harry H. Wright,3 John P. Hussman,4 John R. Gilbert,1 and Margaret A. Pericak-Vance1 1 Duke University Medical Center, Durham, NC North Carolina State University, Raleigh, NC 3 University of South Carolina School of Medicine, Columbia, SC 4 Hussman Foundation, Ellicott City, MD 2 Unlike other complex diseases, the study of autism has been almost exclusively limited to Caucasian families. This study represents a first effort to examine clinical and phenotypic findings in individuals with autism from African American families. Drawing from an ongoing genetic study of autism we compared African American (N ¼ 46, mean age ¼ 118 months) and Caucasian (N ¼ 298, mean age ¼ 105 months) groups on autism symptoms and developmental language symptoms. The African American group showed greater delays in language but did not differ from the Caucasian group on core autism symptoms. These findings, while suggestive of a more severe phenotype, may reflect an ascertainment bias. Nonetheless, we believe that more studies of racial-ethnic groups should be conducted with several goals in mind including strengthening recruiting strategies to include more ethnic-racial groups and more thoughtful evaluation of phenotypic traits. Such considerations will aid greatly in the search for genetic variants in autism. ß 2007 Wiley-Liss, Inc. KEY WORDS: autism; phenotype; African American; racial-ethnic; genetics Please cite this article as follows: Cuccaro ML, Brinkley J, Abramson RK, Hall A, Wright HH, Hussman JP, Gilbert JR, Pericak-Vance MA. 2007. Autism in African American Families: Clinical-Phenotypic Findings. Am J Med Genet Part B 144B:1022–1026. INTRODUCTION Autism is a neurodevelopmental disorder characterized by social-communicative impairments and the presence of repetitive and restricted behaviors [American Psychiatric Association, 2000]. Prevalence studies suggest that autism is more common than originally believed and recent estimates cite a rate of 1/300 [Fombonne, 2003]. While there has been substantial scientific inquiry of clinical, etiologic, and neuro- Grant sponsor: National Institute of Neurological Disorders and Stroke; Grant numbers: P01NS026630, R01NS036768; Grant sponsor: National Alliance of Autism Research. *Correspondence to: Michael L. Cuccaro, Ph.D., Miami Institute for Human Genomics, Miller School of Medicine, University of Miami, Miami, FL. E-mail: firstname.lastname@example.org Received 3 August 2006; Accepted 23 February 2007 DOI 10.1002/ajmg.b.30535 ß 2007 Wiley-Liss, Inc. biological aspects of autism [Santangelo and Tsatsanis, 2005] only a handful of studies have been extended to nonCaucasians. For instance, there have been few studies of prevalence which have included African Americans. Those studies which have indirectly or directly examined prevalence in this group have yielded mixed results. It has been proposed that autism occurs equally across race and culture [Fombonne, 2003]. Support for this hypothesis was found in a large scale population based study of the Atlanta region which reported similar occurrence rates of autism in Caucasian and African American children [Yeargin-Allsopp et al., 2003]. However, two studies noted higher prevalence rates in African American children [Croen et al., 2002; Hillman et al., 2000]. While these results may be a consequence of ascertainment bias (e.g., African American families may only come to services if the child with autism is more severely affected or have difficulties accessing the health care system), at a minimum, prevalence rates in African Americans appear to be at least comparable to those in Caucasians. Studies of autism in African Americans have focused on differences in clinical practices and service utilization [Mandell and Novak, 2005]. For instance, African American children are more likely to be identified with autism based on school records versus clinic based evaluations [YearginAllsopp et al., 2003]. Mandell et al.  report that relative to Caucasian children, African American children are more likely to be diagnosed significantly later. According to Dyches et al. , in a review of multicultural issues in autism, current research has not fully investigated the relationships, if any, between genetic or biological etiologies of autism and race. The autism phenotype, while well defined, is highly variable [Santangelo and Tsatsanis, 2005]. It is believed that this clinical heterogeneity may reflect underlying genetic differences. Again, while autism affects all racial and ethnic groups, a survey of autism genetic studies conducted in the last 10 years in the United States reveals that race is either not mentioned or restricted to families or individuals from Caucasian families [Cantor et al., 2005; Rabionet et al., 2006; Ramoz et al., 2006]. The underlying rationale for studying genetic variations in different racial and ethnic groups is that a complex disorder may have differential liability alleles for different populations; what is learned from one population (e.g., Caucasians) may not transfer seamlessly to another population (e.g., African Americans). There is an abundance of research that points to phenotypic and genotypic differences in African American individuals in complex diseases believed to have a strong genetic component. Genetic differences in African Americans have been identified in cardiovascular disease [Reiner et al., 2005], prostate cancer [Rebbeck, 2005], Alzheimer Disease, and schizophrenia For instance, in Alzheimer Disease, the APOE4 allele causes increased risk among Caucasian subjects, while apolipoprotein D might be a more important risk factor in African Americans [Desai et al., 2003]. Autism in African American Families Studies of the NOTCH4 gene at 6p22-6p24, reveal both a susceptibility haplotype (1725G/25T) for schizophrenia in African Americans [Luo et al., 2004] and excess transmission of the 8 and 13 repeat alleles of the (TAA)n marker in the NOTCH4 gene in African American participants with schizophrenia [Skol et al., 2003]. While such diseases are highly influenced by environmental factors, such results are consistent with the position that risk alleles may have differential effects in different racial groups [Risch et al., 2002]. Phenotype differences have been found in African American versus Caucasian patients in complex genetic disorders including Prader–Willi syndrome on dimensions such as growth, hand and foot lengths, and facial appearance [Hudgins et al., 1998], and Attention Deficit/Hyperactivity Disorder on measures of AD/HD symptom severity [Epstein, 2002]. This study is a preliminary investigation of clinical features of autism in a sample of African American families. The primary objective of this study was to determine if there are meaningful clinical phenotypic differences between the African American and Caucasian autism participants. Clearly, defining phenotypic subgroups in genetic analyses enhances efforts to identify genes in complex disorders such as autism [Coon, 2006]; the study of clinical variation in different populations can inform subgroup development. Identifying clinical differences in different populations represents an approach to subgrouping that has the potential to point to variations at the genetic level. Finally, this study is an important first step in alerting clinicians to the significance of specific symptoms (e.g., severe language deficits) in African American patients where autism may be suspected. Subsequent studies that clarify the relationship, if any, between phenotypic differences and genetic variations may be an aid in developing intervention strategies aimed at specific symptoms not only in African Americans but across all individuals with autism spectrum disorders. METHOD Participants A total of 344 individuals with autism (n ¼ 46 African American; 298 Caucasian) were drawn from a large multi-site study of autism genetics conducted in the southeastern United States. Participants and families were recruited via support groups, advertisements, and clinical and educational settings. Participants were from both multiple incidence (more than one affected individual) and single incidence families. In multiple incidence families only results from one affected individual were used in order to maintain independence of observations. Core inclusion criteria included chronological age between 3 and 21 years of age, a presumptive clinical diagnosis of autism, and expert clinical determination of autism using DSM-IV criteria and supported by the Autism Diagnostic Interview-R [Rutter et al., 2003] or Autism Diagnostic Observation Schedule [Lord et al. 1999]. Exclusion criteria were developmental level <18 months of age, severe sensory problems (e.g., visual impairment or hearing loss), significant motor impairments (e.g., failure to sit by 12 months or walk by 24 months), or identified metabolic, genetic, or progressive neurological disorders based on screening by clinical staff. Race was self-reported by families. Families were defined as African American if one or both parents were of AfricanAmerican descent and they indicated that their child was of African American descent. Three families with children of mixed races were included in the African American sample. The groups were evenly distributed with respect to male: female ratio (M:F AA ¼ 4.7:1, CA ¼ 4.1:1). Table I provides an overview of the demographic characteristics. The groups did not significantly differ on age of onset and age of walking. Also, 1023 TABLE I. Sample Characteristics and Distribution of Functional Language Level (ADI-R q19 Rating) by Race Sample characteristics Number of males Number of females Total participants Mean (SD) age of examb Mean (SD) age walkingb Mean (SD) age of onsetb Mean (SD) age at first wordb,c Mean (SD) age at phrase speechb,c Level of language Functional language Non-functional language African Americana Caucasian American 38 8 46 117.7 (50.6) 12.6 (3.0) 19.2 (8.5) 38.6 (17.6) 59.4 (31.1) 240 58 298 104.7 (47.0) 13.0 (3.5) 19.7 (13.1) 26.7 (15.0) 41.3 (20.0) 23 (50%) 23 (50%) 227 (76%) 71 (24%) a African American sample includes three families of mixed race. Age measured in months. Includes only individuals with functional language. b c while there is a likelihood that the groups were different with respect to age at exam (P ¼ 0.08), this difference was not statistically significant. Procedure All families were recruited through an ongoing autism genetics study. Informed consent was obtained from parents/ caregivers for minor participants. As part of the study, all participants were evaluated using a standard set of diagnostic and clinical assessments. Measurement of Clinical Features Caregivers were administered the Autism Diagnostic Interview-Revised (ADI-R) as part of the ascertainment assessment battery. The ADI-R [Lord et al., 1994, 1997; Rutter et al., 2003] is an investigator-based, semi-structured interview for caregivers that is used to assess diagnostic symptoms of autism. The ADI-R uses a diagnostic algorithm for autism based on DSM-IV diagnostic criteria [American Psychiatric Association, 1994; Filipek et al., 2000] and serves as a clinical research diagnostic standard. Diagnostic algorithms are based on total scores for the Reciprocal Social Interaction, Communication, and Restricted and Repetitive Behaviors or Interests domains using empirically identified item sets. Total scores for these domains are derived based on the presence and severity of behaviors occurring either ever early in development or at any point in development. Individual items constituting the domain scores are coded 0 (not present) to 3 (present and severe). In constructing the algorithm, items scored as 3 are recoded as 2 to reduce undue weight attached to any one item [Rutter et al., 2003]. In addition to ADI-R domain scores, age of first single words (q12), age of first phrases (q13), and overall level of language (q19) were examined. Statistical Analysis Data analyses were designed to compare the African American and Caucasian participants on variables constructed from the ADI-R. All statistical analyses were conducted using SAS 8.02 [SAS Institute, 2000]. The analyses were conducted in stages. In the first stage, the variables of interest were the ADI-R Reciprocal Social Interaction (RSI), Non-Verbal Communication (NVC), Repetitive Behavior (REP) domain scores and Evidence of Abnormalities Before 36 Months of Age (ABN) score. Group differences were examined using MANOVA. The next stage of analysis 1024 Cuccaro et al. examined only participants who had functional language as indicated by a score of 0 on ADI-R question 19 (Overall Level of Language). A score of 0 is defined as ‘‘functional use of spontaneous, echoed or stereotyped language that, on a daily basis, involves phrases of three words or more that at least sometimes include a verb and is comprehensible to other people.’’ The groups were compared on the Verbal Communication domain using ANOVA as well as chi-square analyses to examine differences on six individual questions that compose the Verbal Communication domain. The final stage of analysis compared the African American and Caucasian participants on ADI-R measures of language development— age of first single words (q12) and age of first phrases (q13). These data were analyzed using ANOVA while accounting for overall level of language (ADI-R question 19). This step was necessary as the two groups differed with respect to level of functional language (50% of the AA sample vs. 76% of CA had a rating of 0 on the overall level of language item of 0, P ¼ 0.0002; See Table I). The final portion of the analysis examined differences in language development between the African American and Caucasian groups using ADI-R questions 12 and 13 (age at first words and age at first phrases) while accounting for overall level of language (ADI-R item 19). This strategy was necessary as the groups differed with respect to functional language level (as noted earlier with P-value of 0.0002). Table III shows ANOVA results that model questions age at first words (ADI-R q12) and age at phrase speech (ADI-R q13) using both dichotomized functional language (responses from question 19) and race. Both models are significant (Q12 F(2, 299) ¼ 10.7, P < 0.0001; Q13 F(2, 267) ¼ 11.08, P < 0.0001). Type 1 analyses show that in both models there are significant race effects in the development of first words and first phrases, even after accounting for the differences in the level of language. As with the individual language items, there are variations in sample size due to coding conventions. There were no significant interactions between race and level of language for either of the above models (results not shown). RESULTS DISCUSSION MANOVA on the domain scores (RSI, NVC, REP, ABN), using a single test statistic (Wilks’ Lambda) resulted in a Pvalue of 0.19 indicating no significant differences between Caucasian and African American scores on the ADI-R domain scores. We then restricted our analyses only to individuals with functional language as indicated by a score of 0 on NVC overall level of language (N ¼ 250; Caucasian N ¼ 227; African American N ¼ 23). Repeating the MANOVA for the RSI, NCV, REP, and ABN domains yielded similar results (P ¼ 0.20). Given our interest in the language functioning of the groups, we examined the Verbal domain separately in this second group of individuals and noted a marginally significant result when race was used to model the ADI-R Verbal domain score (F(1,248) ¼ 3.01, P ¼ 0.08). Despite the lack of statistically significant findings we explored the marginally significant ANOVA finding—examining the six ADI-R items (item numbers 16, 18, 20, 22, 23, 24) that sum to provide the ADI— Verbal Communication domain score. Individual item scores were recoded into low (0–1) response versus high (2–3) response and chi-square tests were used to determine if there were any differences between the two groups on these six questions (Table II). No significant differences were apparent in the distribution of scores between the two groups as P-values ranged from 0.06 to 0.46. Note that variations in sample size between items are the result of coding responses such as not appropriate or not known on select items. Since no test met the threshold of significance, retrospective power calculations of likelihood ratio chi-square tests were calculated (see Table II) using SAS [SAS Institute, 2000]. As seen in Table II it is clear that our power to detect significant effects in these analyses were severely limited by our sample size suggesting that a larger sample size of African Americans is needed. Genotypic and phenotypic differences in African American participants have been found in complex neuropsychiatric diseases such as schizophrenia, AD/HD, and Alzheimer’s disease. This is the first study of phenotypic features in African American participants with autism. Based on detailed caregiver report over a variety of symptom domains from the ADI-R, the African American participants showed significantly later acquisition of first words and phrase speech. This developmental difference stands in contrast to no differences in the mean scores on the ADI-R social and repetitive behavior domain scores. Even the ADI-R language composite scores were not significantly different. However, the presence of functional language was very different in our groups as was the age at acquisition suggesting the possibility of a more severe language phenotype in our African American participants. Phenotypic differences in African American participants with autism may reflect underlying differences in genetic susceptibility unique to this population. The parent reported presence of a more severe language phenotype in the African American group assumes greater significance in light of studies that suggest acquisition of language may index genetically meaningful subgroups [Tager-Flusberg and Joseph, 2003; Alarcon et al., 2005]. While no population specific genetic differences have been identified in African American families, Collins et al.  have found evidence of significant association to autism in SNPS in GABRA4 and GABRB1 in African-Americans (rs2280073, P ¼ 0.0287, and rs16859788, P ¼ 0.0253) confirming results previously detected in a Caucasian dataset. Efforts to determine if different genetic variations are present in African American autism families can use information about phenotypic differences to explore this further. For instance, selecting both TABLE II. Distribution of Item Scores on ADI-R Language Items for the African American and Caucasian Participants Caucasian ADI—R item (#) Social vocalization/chat (Q16) Reciprocal conversation (Q20) Stereotype utterances/delayed echolalia (Q18) Inappropriate questions/statements (Q22) Pronominal reversal (Q23) Neologisms/idiosyncratic language (Q24) African American Low (0–1) High (2–3) Low (0–1) High (2–3) Chi-square P-value Retrospective power 162 110 92 157 115 194 60 113 130 63 97 30 12 6 7 17 11 20 9 15 14 3 6 1 2.20 3.43 0.53 1.91 0.71 1.61 0.14 0.06 0.46 0.17 0.40 0.20 0.32 0.46 0.11 0.28 0.13 0.25 Autism in African American Families TABLE III. Comparisons of Age at First Words (ADI-R q12) and Age at Phrase Speech (ADI-R q13) by Race Accounting for Differences in Language Level Sum of squares Q12 (age at first words) Model Level of language Race Error Total Q13 (age at phrase speech) Model Level of language Race Error Total a F-statistic P-value 5,394.45a 2,492.40a 14.63 6.76 110,267.99 118,156.84 0.0002 0.0098 4,598.77a 6,029.97a 9.59 12.57 128,073.56 138,702.30 0.0022 0.0005 Type I sum of squares. African American and Caucasian families with severe language impairments one could examine whether there are genetic differences in these subsets. Other explanations for phenotypic differences should be entertained. An important consideration is whether the observed differences in language severity reflect a clinical or ascertainment bias (i.e., for the African American participants only those with the more severe phenotype came to the attention of clinical providers or enrolled in the study). Further, it is not known if this bias is unique to the current study or more characteristic of African American autism families. In each case, the bias explanation holds that only the African American families with the more severely affected children were participants in this study. The bias could be the result of differences in referral patterns or help-seeking and as a result only those families in which children have severe impairments make their way to medical professionals. A related possibility is that severity level of the child influences the willingness of caregivers to participate in genetic studies of autism. While we are not able to discount such a bias, it is important to note that the severity effect was not evident across all domains of functioning. Further bias may be related to access to services that can ameliorate early language problems. As noted above we did not find that the African American participants had greater impairments in social and repetitive domains—independent features of the autism phenotype. Also, while several studies have suggested that African American individuals with autism are diagnosed later than Caucasian individuals we found that the groups did not differ with respect to age at which the parents recognized the initial symptoms. While this is less tied into the discussion at hand, it is an important point to consider in the clinical realm (i.e., if parents are noting these symptoms at about the same age, why are the African American participants being identified much later). Also, it would be interesting to examine the type of symptoms that first arouse interest in the groups. A third possibility is that the African American group demonstrates a more severe language phenotype, that is, independent of autism (i.e., not autism specific). For instance, in the case of language, it may be that the delay in language acquisition occurs in the African American population and we are capturing these same phenomena in our African American autism participants. While all hypotheses should be entertained, there is no evidence to suggest that there are differences among the various populations with respect to acquisition of single words or phrase speech. To the contrary, these milestones seem to be universal and occur with 1025 regularity among the various ethnic-racial groups. Further, in a study of preschoolers with speech delay, the factors such as family history and maternal education were significant risk predictors while race was not associated with increased risk [Campbell et al., 2003]. However, recent evidence suggests that there may be differences in how parents report on language skills as a function of SES and race [Roberts et al., 1999]. Further, there is evidence that language experiences vary according to SES. For instance, Hoff and colleagues have reported in a series of studies that the language experiences of children significantly impact language development [HoffGinsberg, 1991; Hoff, 2003]. Specifically, there appear to be SES related differences in the speech that children hear which in turn affects their language skills [Hoff, 2003]. This strong relationship between SES, language experiences, and language development underscores the potential significance of SES in any studies of language in autism spectrum disorders. As this study represents the first effort to examine clinical phenotypic features in African American individuals with autism it should be considered preliminary. Certainly such results warrant careful examination of other non-Caucasian populations as well as larger scale studies of African American individuals with autism. Given the preliminary nature of the work, it is important to note that are a number of limitations which should be recognized and incorporated into future studies as appropriate. First, the absence of control participants reduces our confidence in whether the findings— particularly with respect to language—are autism specific. A second concern is the limited assessment battery. The data reported are all derived from parent/caregiver report on behavior. Direct assessment of language and other domains may serve to further elucidate the autism phenotype in African American participants. Further, ADOS data was available for only a limited number of participants in the African American sample. This information would have allowed us to have an index of behavior, including language, independent of parent report. A third limitation is the lack of information about SES which may also influence symptom presentation—particularly with respect to reporting on language abilities. A final limitation concerns age at diagnosis of autism which has been shown to be significantly later for African American children [Mandell et al., 2005]. While we did not directly compare the two groups on age at diagnosis, we did find that the age at which parents had first concerns about development was later for the African American children with autism although this difference was not statistically significant (P ¼ 0.74). Recruiting families from different racial-ethnic groups for autism genetics research is crucial to our understanding of the complexity of this disorder. It is common practice to separate racial-ethnic groups based on the premise that there are different genetic or environmental causes in different populations. While preliminary, the current study raises questions about how to integrate different populations into genetic studies of autism. Our findings suggest that those studies which include mixed populations may be prone to difficulties due to ascertainment related influences. For instance, constructing subsets based on phenotypes that are potentially influenced by ascertainment differences may be confounding (e.g., only African American families whose children have severe language or behavior problems may access services or enroll) and introduce a nonrandom element into the study. Correcting for such a situation may be as simple as analyzing the groups separately. However, it does speak to the importance of seeking to understand and explain the full range of contributions to phenotype. One approach which will enhance understanding of environmental and genetic contributions to clinical phenomena in autism (e.g., language) is a thorough examination of phenotype and course over time. This could be 1026 Cuccaro et al. accomplished by careful collection of historical data as well as attention to prospective methods to examine the trajectory of clinical variables. For the most part, the study of phenotype has focused on single points in time with little consideration of the persistence of traits over time. In conjunction with a strengthening of recruiting strategies to ensure diverse representation we can dissect the contributions to phenotype and discern more clearly whether phenotypic differences are relevant to genetic variation. ACKNOWLEDGMENTS We thank the patients with autism, and family members who agreed to participate in this study, as well as the personnel of the Center for Human Genetics at Duke University Medical Center where this research was conducted. We also thank Dr. Robert Delong and Dr. Gordon Worley of Duke University Medical Center for referring patients and their families to the study. This research was supported in part by grants from the National Institute of Neurological Disorders and Stroke (P01NS026630, R01NS036768); funding from the National Alliance of Autism Research; and a gift from the Hussman Foundation. The research conducted in this study complies with current U.S. laws. Hillman RE, Kanafani N, Takahashi TN, Miles JH. 2000. Prevalence of autism in Missouri: Changing trends and the effect of a comprehensive state autism project. Mo Med 97:159–163. Hoff E. 2003. The specificity of environmental influence: Socioeconomic status affects early vocabulary development via maternal speech. Child Dev 74:1368–1378. Hoff-Ginsberg E. 1991. Mother-child conversation in different social classes and communicative settings. Child Dev 62:782–796. Hudgins L, Geer JS, Cassidy SB. 1998. Phenotypic differerencss in African Americans with Prader-Willi syndrome. Genet Med 1:49–51. Lord C, Rutter M, Le Couteur A. 1994. Autism diagnostic interview-revised: A revised version of a diagnostic interview for caregivers of individuals with possible pervasive developmental disorders. J Autism Dev Disord 24:659–685. Lord C, Pickles A, McLennan J, Rutter M, Bregman J, Folstein S, Fombonne E, Leboyer M, Minshew N. 1997. Diagnosing autism: Analyses of data from the Autism Diagnostic Interview. J Autism Dev Disord 27:501–517. Lord C, Rutter M, DiLavore P, Risi S. 1999. Autism Diagnostic Observation Schedule-WPS (WPS Edition). Los Angeles: Western Psychological Association. Luo X, Klempan TA, Lappalainen J, Rosenheck RA, Charney DS, Erdos J, van Kammen DP, Kranzler HR, Kennedy JL, Gelernter J. 2004. NOTCH4 gene haplotype is associated with schizophrenia in African Americans. Biol Psychiatry 55:112–117. Mandell DS, Novak M. 2005. The role of culture in families’ treatment decisions for children with autism spectrum disorders. Ment Retard Dev Disabil Res Rev 11:110–115. REFERENCES Mandell DS, Listerud J, Levy SE, Pinto-Martin JA. 2002. Race differences in the age at diagnosis among medicaid-eligible children with autism. J Am Acad Child Adolesc Psychiatry 41:1447–1453. Alarcon M, Yonan AL, Gilliam TC, Cantor RM, Geschwind DH. 2005. Quantitative genome scan and ordered-subsets Analysis of autism endophenotypes support language QTLs. Mol Psychiatry 10:747–757. Mandell DS, Novak MM, Zubritsky CD. 2005. Factors associated with age of diagnosis among children with autism spectrum disorders. Pediatrics 116:1480–1486. 1994. American Psychiatric Association. Diagnostic and statistical manual of mental disorders (DSM-IV). Vol. IV. Washington, DC: American Psychiatric Press, Inc. Rabionet R, McCauley Ph DJ, Jaworski MPHJ, Ashley-Koch AE, Martin ER, Sutcliffe JS, Haines JL, DeLong GR, Abramson RK, Wright HH, Cuccaro ML, Gilbert JR, Pericak-Vance MA. 2006. Lack of association between autism and SLC25 A12. Am J Psychiatry 163:929–931. 2000. American Psychiatric Association. Diagnostic and statistical manual of mental disorders (DSM-IV), text revision. Washington, DC: American Psychiatric Press, Inc. Campbell TF, Dollaghan CA, Rockette HE, Paradise JL, Feldman HM, Shriberg LD, Sabo DL, Kurs-Lasky M. 2003. Risk factors for speech delay of unknown origin in 3-year-old children. Child Dev 74:346–357. Cantor RM, Kono N, Duvall JA, Alvarez-Retuerto A, Stone JL, Alarcon M, Nelson SF, Geschwind DH. 2005. Replication of autism linkage: Finemapping peak at 17q21. Am J Hum Genet 76:1050–1056. Collins AL, Ma D, Whitehead PL, Martin ER, Wright HH, Abramson RK, Hussman JP, Haines JL, Cuccaro ML, Gilbert JR, Pericak-Vance MA. 2006. Investigation of autism and GABA receptor subunit genes in multiple ethnic groups. Neurogenetics 7:167–174. Coon H. 2006. Current perspectives on the genetic analysis of autism. Am J Med Genet Part C 142:24–32. Croen LA, Grether JK, Selvin S. 2002. Descriptive epidemiology of autism in a California population: Who is at risk? J Autism Dev Disord 32:217–224. Ramoz N, Reichert JG, Corwin TE, Smith CJ, Silverman JM, Hollander E, Buxbaum JD. 2006. Lack of evidence for association of the serotonin transporter gene SLC6A4 with autism. Biol Psychiatry 60:186–191. Rebbeck TR. 2005. Genetics, disparities, and prostate cancer. LDI Issue Brief 10:1–4. Reiner AP, Ziv E, Lind DL, Nievergelt CM, Schork NJ, Cummings SR, Phong A, Burchard EG, Harris TB, Psaty BM, Kwok PY. 2005. Population structure, admixture, and aging-related phenotypes in African American adults: The Cardiovascular Health Study. Am J Hum Genet 76:463–477. Risch N, Burchard E, Ziv E, Tang H. 2002. Categorization of humans in biomedical research: Genes, race and disease. Genome Biol 3:1–12, comment2007. Roberts JE, Burchinal M, Durham M. 1999. Parents’ report of vocabulary and grammatical development of African American preschoolers: Child and environmental associations. Child Dev 70:92–106. Desai PP, Hendrie HC, Evans RM, Murrell JR, DeKosky ST, Kamboh MI. 2003. Genetic variation in apolipoprotein D affects the risk of Alzheimer disease in African-Americans. Am J Med Genet Part B 116B: 98–101. Rutter M, LeCouteur A, Lord C. 2003. Autism Diagnostic Interview, Revised (ADI-R). Los Angeles, CA: Western Psychological Services. Dyches TT, Wilder LK, Sudweeks RR, Obiakor FE, Algozzine B. 2004. Multicultural issues in autism. J Autism Dev Disord 34:211–222. 2000. SAS Institute. SAS/STAT User’s Guide (Version 8). Vol. 1 and 2. 4th edition. Cary, NC: Author. Epstein FH. 2002. Spironolactone and ACE inhibition in chronic renal failure. N Engl J Med 346:456. Skol AD, Young KA, Tsuang DW, Faraone SV, Haverstock SL, Bingham S, Prabhudesai S, Mena F, Menon AS, Yu CE, Rundell P, Pepple J, Sauter F, Baldwin C, Weiss D, Collins J, Keith T, Boehnke M, Schellenberg GD, Tsuang MT. 2003. Modest evidence for linkage and possible confirmation of association between NOTCH4 and schizophrenia in a large Veterans Affairs Cooperative Study sample. Am J Med Genet Part B 118B:8–15. Filipek PA, Accardo PJ, Ashwal S, Baranek GT, Cook EH Jr, Dawson G, Gordon B, Gravel JS, Johnson CP, Kallen RJ, Levy SE, Minshew NJ, Ozonoff S, Prizant BM, Rapin I, Rogers SJ, Stone WL, Teplin SW, Tuchman RF, Volkmar FR. 2000. Practice parameter: Screening and diagnosis of autism: Report of the Quality Standards Subcommittee of the American Academy of Neurology and the Child Neurology Society. Neurology 55:468–479. Fombonne E. 2003. Epidemiological surveys of autism and other pervasive developmental disorders: An update. J Autism Dev Disord 33:365–382. Santangelo SL, Tsatsanis K. 2005. What is known about autism: Genes, brain, and behavior. Am J Pharmacogenomics 5:71–92. Tager-Flusberg H, Joseph RM. 2003. Identifying neurocognitive phenotypes in autism. Philos Trans R Soc Lond B Biol Sci 358:303–314. Yeargin-Allsopp M, Rice C, Karapurkar T, Doernberg N, Boyle C, Murphy C. 2003. Prevalence of autism in a US metropolitan area. JAMA 289:49–55.