Scholarly article on topic 'Emerging cognitive profiles in high-risk infants with and without autism spectrum disorder'

Emerging cognitive profiles in high-risk infants with and without autism spectrum disorder Academic research paper on "Psychology"

CC BY-NC-ND
0
0
Share paper
OECD Field of science
Keywords
{"High-risk siblings" / "Cognitive development" / Infants / Toddlers / "Developmental trajectories" / ASD}

Abstract of research paper on Psychology, author of scientific article — A. Jessica Brian, C. Roncadin, E. Duku, S.E. Bryson, I.M. Smith, et al.

Abstract This paper examined early developmental trajectories in a large, longitudinal sample at high-risk for ASD (‘HR’) and low-risk (‘LR’) controls, and the association of trajectories with 3-year diagnosis. Developmental assessments were conducted at 6, 12, 24 months, and 3 years, with blinded “clinical best-estimate” expert diagnosis at age 3. HR infants were enrolled based only on familial risk. LR infants, from community sources, had no first- or second-degree ASD relatives. All infants were born at 36–42 weeks, weighing ≥2500g, with no identifiable neurological, genetic, or severe sensory/motor disorders. Analytic phase I: semi-parametric group-based modeling to identify distinct developmental trajectories (n =680; 487 HR; 193 LR); phase II: Trajectory membership in relation to 3-year diagnosis (n =424; 310 HR; 114 LR). Three distinct trajectories emerged (1) inclining; (2) stable-average; (3) declining; trajectory membership predicted diagnosis (χ 2 =99.40; p <.001). Most ASD cases were in stable-average (50.6%) or declining trajectories (33.8%); most non-ASD-HR infants were in inclining (51.9%) or stable-average (40.3%) trajectories. The majority of LR controls were in the inclining trajectory (78.9%). Within the declining trajectory, over half had ASD (57.8%), but 40% were non-ASD-HR infants. Declining/plateauing raw scores were associated with, but not exclusive to, ASD. Findings underscore the importance of monitoring the emergence of ASD symptoms and overall development in high-risk children. Evidence of developmental slowing or decline may be associated not only with ASD, but with other suboptimal outcomes, warranting careful clinical follow-up.

Academic research paper on topic "Emerging cognitive profiles in high-risk infants with and without autism spectrum disorder"

Contents lists available at ScienceDirect

Research in Autism Spectrum Disorders

Journal homepage http://ees.elsevier.com/RASD/default.asp

Emerging cognitive profiles in high-risk infants with and without autism spectrum disorder

A. Jessica Brian a'b'*, C. Roncadinc,d, E. Dukue, S.E. Brysonf, I.M. Smithf, W. Roberts g, P. Szatmarih, I. Drmici, L. Zwaigenbaumj

a Autism Research Centre, Holland Bloorview Kids Rehabilitation Hospital, Dept. Paediatrics, University of Toronto, Canada b Autism Research Unit (ARU), SickKids, School of Graduate Studies, University of Toronto, Toronto, Canada cAutism Services, Kinark Child and Family Services, Markham, ON, Canada d School of Graduate Studies, University of Toronto, Toronto, Canada

e Dept. Psychiatry and Behavioural Neurosciences, McMaster University, Hamilton, ON, Canada f!WK Health Centre/Dalhousie University, Halifax, NS, Canada g Autism Research Unit/SickKids and Dept. Paediatrics, University of Toronto, Canada h Hospital for Sick Children and Centre for Addiction and Mental Health Toronto, Canada i ARU/SickKids and Holland Bloorview, Toronto, Canada

j Autism Research Centre, Glenrose Rehabilitation Hospital, Department of Pediatrics, University of Alberta, Edmonton, AB, Canada

CrossMark

ARTICLE INFO

ABSTRACT

Article history:

Received 5 June 2014

Received in revised form 25 July 2014

Accepted 27 July 2014

Available online 14 September 2014

Keywords: High-risk siblings Cognitive development Infants Toddlers

Developmental trajectories ASD

This paper examined early developmental trajectories in a large, longitudinal sample at high-risk for ASD ('HR')and low-risk ('LR') controls, and the association of trajectories with 3-year diagnosis. Developmental assessments were conducted at 6, 12, 24 months, and 3 years, with blinded ''clinical best-estimate'' expert diagnosis at age 3. HR infants were enrolled based only on familial risk. LR infants, from community sources, had no first- or second-degree ASD relatives. All infants were born at 36-42 weeks, weighing >2500 g, with no identifiable neurological, genetic, or severe sensory/motor disorders. Analytic phase I: semi-parametric group-based modeling to identify distinct developmental trajectories (n = 680; 487 HR; 193 LR); phase II: Trajectory membership in relation to 3-year diagnosis (n = 424; 310 HR; 114 LR). Three distinct trajectories emerged (1) inclining; (2) stable-average; (3) declining; trajectory membership predicted diagnosis (x2 = 99.40; p < .001). Most ASD cases were in stable-average (50.6%) or declining trajectories (33.8%); most non-ASD-HR infants were in inclining (51.9%) or stable-average (40.3%) trajectories. The majority of LR controls were in the inclining trajectory (78.9%). Within the declining trajectory, over half had ASD (57.8%), but 40% were non-ASD-HR infants. Declining/ plateauing raw scores were associated with, but not exclusive to, ASD. Findings underscore the importance of monitoring the emergence of ASD symptoms and overall development in high-risk children. Evidence of developmental slowing or decline may be associated not only with ASD, but with other suboptimal outcomes, warranting careful clinical follow-up. © 2014 The Authors. Published by Elsevier Ltd. This is an open access article under the CC BY-NC-ND license (http://creativecommons.org/licenses/by-nc-nd/3.0/).

* Corresponding author at: Bloorview Research Institute, University of Toronto, 150 Kilgour Rd., Toronto, ON M4G 1R8, Canada. Tel.: +1 416 425 6220x3716; fax: +1 416 422 7045.

E-mail addresses: jbrian@hollandbloorview.ca (A.J. Brian), caroline.roncadin@kinark.on.ca (C. Roncadin), Duku@mcmaster.ca (E. Duku), Susan.bryson@iwk.nshealth.ca (S.E. Bryson), Isabel.smith@iwk.nshealth.ca (I.M. Smith), Wendy.roberts@sickkids.ca (W. Roberts), peter.szatmari@utoronto.ca (P. Szatmari), idrmic@hollandbloorview.ca (I. Drmic), lonnie.zwaigenbaum@albertahealthservices.ca (L. Zwaigenbaum).

http://dx.doi.org/10.1016/j.rasd.2014.07.021

1750-9467/© 2014 The Authors. Published by Elsevier Ltd. This is an open access article under the CC BY-NC-ND license (http://creativecommons.org/ licenses/by-nc-nd/3.0/).

1. Introduction

Autism spectrum disorder (ASD) is a neurodevelopmental disorder characterized by social-communication deficits and repetitive/restricted behavior (APA, 2013). Recent prevalence estimates of 1/68 (CDC, 2014), with recurrence of close to 20% in younger siblings (Ozonoff et al., 2011), underscore the need for careful monitoring of this high-risk population. Considerable heterogeneity exists in ASD, with variability not only in ASD symptom presentation, but also in intellectual development. Intellectual functioning in ASD varies along a continuum from severe impairment to well above average and can significantly impact daily functioning, school placement, and prognosis (Harris & Handleman, 2000; Howlin, Goode, Hutton, & Rutter, 2000). Intellectual deficits are reported to occur in 30-50% of cases (Chakrabarti & Fombonne, 2005; CDC, 2014), but characterization of intellectual disability in ASD and cross-study comparisons are complicated by sample heterogeneity and different conceptualizations/measures of "intelligence." Moreover, despite the assumption that intelligence/IQ is stable, strong evidence of variability over time exists, even into adolescence, in typical development (Ramsden et al., 2011). The issue of IQ stability may be particularly relevant to special populations, such as those with low intellectual functioning (Whitaker & Taylor, 2008), dyslexia (Ingesson, 2006), low birth weight (Mortensen, Andressen, Krusse, & Sanders, 2003), or ASD, in which variability is a particular consideration under age three (Lord & Schopler, 1989a, 1989b). While early improvement in IQ(i.e., from 2 to 4 years) predicts better outcomes in ASD (Sutera et al., 2007; Kelley, Naigles, & Fein, 2010), questions persist as to whether differential trajectories of intellectual functioning in the first years of life in infants at high risk for ASD are associated with different outcomes, or whether declining intellectual performance mirrors other forms of regression, reported in 15-50% of the ASD population (Barger, Campbell, & McDonough, 2013; Ekinci, Arman, Melek, Bez, & Berkem, 2012; Stefanatos, 2008; Yirmiya & Charman, 2010).

Longitudinal studies of younger siblings of children with ASD ('high-risk' infants; 'HR') reveal reduced rates of skill acquisition on the Mullen Scales of Early Learning (MSEL; Mullen, 1995) after age 6-12 months in some infants later diagnosed with ASD (Bryson et al., 2007; Landa, Holman, & Garrett-Mayer, 2007; Zwaigenbaum et al., 2005). However, existing findings are limited by small samples, thus precluding the examination of heterogeneity. Considerable variability is anticipated in this high-risk group, given the range in intellectual functioning at the time of diagnosis even in young children, and based on our earlier case series report in which declining trajectories and severe cognitive impairment at 2-3 years were associated with an ASD diagnosis and earlier symptom onset (Bryson et al., 2007). It remains unclear whether atypical trajectories are specific to ASD, as individual differences have been obscured by group designs used to date (e.g., comparing trajectories based on outcomes, rather than examining the variation in trajectories across the high-risk group as a whole).

A recent latent class analysis of intellectual development examined domain scores on the MSEL (Mullen, 1995) in 204 HR infants aged 6-36 months. A three- and a four-class model emerged, each with strong classification quality (.92 and .89 entropy, respectively), the latter favored by the authors (Landa, Gross, Stuart, & Bauman, 2012). Infants with ASD outcomes were over-represented in a 'developmental slowing' class that was highly specific to ASD (>90% of children in this class had ASD). However, over half of those with ASD fell into one of three other classes (motor/receptive language delays, or average/ accelerating to above-average language skills). A related study (Landa, Gross, Stuart, & Faherty, 2013) that included the original sample plus a small group of low-risk (LR) controls (n = 31) revealed skill loss in both language domains of the MSEL in 24% of the ASD participants (other domains were not examined). Four additional "non-ASD" cases had skill loss, but it is not clear whether these were from the HR or LR group because non-ASD groups were collapsed.

These studies provide initial evidence of distinct developmental trajectories in HR infants that are associated with ASD. The current paper extends these findings and is distinguished by the use of overall cognitive scores to determine trajectories, the largest sample to date in this type of investigation, and the preservation of the distinction between non-ASD HR and LR participants, supported by evidence that non-ASD HR siblings represent a unique group (Ben-Yizhak et al., 2011; Bishop, Mayberry, Wong, Maley, & Hallmayer, 2006; Georgiades et al., 2013; Messinger et al., 2013; see Drumm & Brian, 2013 for a recent review).

2. Methods

2.1. Participants

Participants were drawn from a longitudinal, prospective study of (HR) younger siblings of children with ASD (Zwaigenbaum et al., 2005), recruited through four major Canadian ASD diagnostic centers and community physicians. HR infants were enrolled at 6 or 12 months of age based only on familial risk. LR infants, recruited through community sources, had no first- or second-degree relatives with ASD. All infants were born at 36-42 weeks gestation, weighing >2500 g; none had identifiable neurological or genetic disorders, or severe sensory/motor impairments. All participants with relevant data from the larger study were included. Data were analyzed in two phases to maximize the use of available data; Phase I: n = 680; 487 HR(267 males); 193 LR(105 males); and Phase II: n = 424 infants (310 HR; 114 LR) followed to at least 3 years of age through our ongoing prospective study (Zwaigenbaum et al., 2005).

2.2. Procedures

Participants were assessed at 6,12, 24 months and 3 years of age using the Mullen Scales of Early Learning (MSEL; Mullen, 1995). Data were not available at all time points for all participants (see Table 1). At age 3, an independent, ''clinical

Table 1

Mean age and Mullen scales of early learning-early learning composite standard score by diagnostic outcome group.

ASD-HR (a) Non-ASD-HR (b) LR(c) ANOVA (MSEL) Post-hoc (MSEL)

Mean (SD) Age (mos) Mean (SD) MSEL Mean (SD) Age (mos) Mean (SD) MSEL Mean (SD) Age (mos) Mean (SD) MSEL Omnibus F (p value) Significant contrasts

6-month assessment 12-month assessment 6.36 (.66) n = 28 12.34 (.56) n = 63 94.79 (9.25) 98.52 (13.99) 6.48 (.74) n = 99 12.44 (.66) n =185 96.80 (11.80) 105.89 (14.99) 6.56 (.60) n = 48 12.30 (.50) n = 94 101.88 (10.26) 111.00 (11.88) 4.73 (p = .010) 14.95 (p < .001) All N.S.* a<b<c ps < .05 a < b < c p's < .001 a<b<c p's < .001

24-month assessment 24.53 (.86) n = 65 86.63 (20.49) 24.55 (.99) n =196 106.65 (17.57) 24.67 (1.06) n = 104 120.07 (15.89) 71.64 (p < .001)

3-year assessment 38.13 (2.75) n = 71 87.03 (24.24) 38.69 (3.16) n = 226 109.50 (18.04) 38.91 (3.40) n = 112 120.42 (15.52) 70.32 (p < .001)

Note. ASD-HR: high-risk infants with ASD diagnosis at age 3; non-ASD-HR: high-risk infants with no diagnosis of ASD; LR: low-risk infants with no diagnosis of ASD; MSEL: Mullen scales of early learning-early 8 learning composite; mos: months. 22

* At 6 months, two contrasts approached significance (with Bonferroni correction): a < c (p = .023) and b < c (p = .029). 2

best-estimate'' diagnosis was made by an expert clinician (psychologist, developmental pediatrician, or child psychiatrist) blind to previous assessments, as described by Zwaigenbaum et al. (2012). ASD diagnoses were assigned using DSM-IV-TR (APA, 2000) criteria, based on information from research-reliable ADI-R (Lord, Rutter, & LeCouteur, 1994) and ADOS (Lord, Rutter, DiLavore, & Risi, 1999; Lord, Risi, & Lambrecht, 2000) administrations. Consistent with best practice (Baird, Douglas, & Murphy, 2011; Risi et al., 2006; Ventola et al., 2006), some participants received a clinical diagnosis of ASD despite sub-threshold ADOS and/or ADI-R scores if they met DSM-IV-TR criteria based on the expert's review of all data from the 3-year assessment. Conversely, participants exceeding threshold on these measures were not necessarily given ASD diagnoses if the expert clinician did not think a diagnosis was warranted. Research ethics approval was provided for all participating institutions, and informed consent was obtained from parents/primary caregivers.

2.3. Measures

The Mullen Scales of Early Learning (MSEL; Mullen, 1995) is a standardized assessment of five developmental domains from 0 to 68 months: Gross Motor (GM), Fine Motor (FM), Visual Reception (VR), Receptive (RL) and Expressive Language (EL). The Early Learning Composite (ELC) is a standard score (mean = 100, SD = 15) representing overall cognitive ability, derived from the latter four subscales. Good internal consistency and test-retest stability are reported (Mullen, 1995).

The Autism Diagnostic Interview-Revised (ADI-R; Lord et al., 1994) is an investigator-directed interview that elicits information relevant to a DSM-IV-TR (APA, 2000) autism diagnosis. Inter-rater reliability is excellent, and the ADI-R discriminates well between autism and other developmental disabilities (Lord et al., 1994; Risi et al., 2006).

The Autism Diagnostic Observation Schedule (ADOS; Lord et al., 1999, 2000) is a standardized, semi-structured direct observation of social-communication, play and behavior with excellent inter-rater reliability and good stability for classification above age two (Lord et al., 2006).

2.4. Analytic plan

Phase I: Semi-parametric group-based modeling (SAS PROC TRAJ) was used to identify distinct developmental trajectories in MSEL-ELC from age 6 months to 3 years. This statistical approach identifies latent sub-populations with common characteristics based on the observed longitudinal patterns in a sample. Per standard practice, we included all cases (i.e., not just those with 3-year outcomes) to improve generalizability of derived trajectories to the entire sample (Nagin, 1999).

Phase II: Subsequent analyses focused on the subset of cases with 3-year outcome data using Chi-square to examine relations between trajectory membership and diagnosis (Fisher's Exact Test for cell counts <5). Univariate ANOVA tested group differences in demographic variables and MSEL-ELC.

3. Results

3.1. Phase I analysis

3.1.1. MSEL-ELC trajectories

Semi-parametric group-based modeling, using our combined HR and LR sample (N = 680), yielded a 3-group solution that provided optimal fit to variation in trajectories (see Fig. 1). Trajectory 1 ('Inclining': n = 359; 211 HR, 148 LR) was characterized by performance beginning within average limits and increasing to >1 standard deviation above the standardization mean by age 2, with a positive linear slope (1.94, p < .001). Trajectory 2 ('Stable-Average': n = 254; 211 HR, 43 LR) entailed consistently average performance across time (linear slope = .27, p = .34). Trajectory 3 ('Declining') involved a marked decline from average to 2 SD below average by age 2, persisting to age 3 (n = 67; 65 HR, 2 LR; linear slope = -2.31, p < .001). Note that trajectory names reflect patterns in standard scores; thus, e.g., a 'declining' pattern reflects a slowing of developmental progress relative to age-based norms. Pairwise comparisons between trajectory groups for the estimates of intercept, linear slope, and quadratic slope revealed significant differences for all comparisons (all p's < .01) with the exception of the intercepts for trajectories 1 and 2 (p = .13). The average trajectory of the group as a whole can be characterized with intercept = 96.74 (standard error; SE=1.59), linear slope = .63 (SE = .16) and quadratic slope = -.01 (SE = .003), indicating significant variation around the intercepts and slopes.

3.1.2. Validity and distribution of ELC scores across trajectories

Given the high probability of sub-domain scatter, particularly in our HR sample, the validity of using the ELC (rather than domain scores) was explored by examining the presence of inter-domain scatter (i.e., domain t-scores > 2 SD apart) at any time point for all cases. Chi-square analyses revealed no association between trajectory membership and proportion of cases with sub-domain scatter at any time point (x2 = 1.10, .22,3.3, and .85 at 6,12,24 months, and 3 years, respectively, p's range: .19-.90). As a rough index of the proportion of cases falling within the range of ''intellectual disability'', we also examined the rate of ELC scores < 70, revealing no cases at 6 months, 1.8% of cases at 12 months, 6% of cases at 24 months, and 6.3% of cases at age 3. Significant associations emerged between proportion of cases with ELC < 70 and trajectory membership at 12 and 24 months, and at 3 years, s (Fisher's Exact = 43.82, x2 = 209.10, and x2 = 208.67, respectively, all p's < .001). At 12 months, all cases (n = 10) with ELC < 70 were from the Declining trajectory; at 24 months and 3 years, the vast majority of cases with

Fig. 1. Three trajectory groups for Mullen scales of early learning-early learning composite (MSEL-ELC) derived from semi-parametric group-based modeling (n = 680). Dashed lines depict upper and lower limits of 95% confidence intervals.

ELC < 70 were also from the Declining trajectory (29/30, and 25/27, respectively). This analysis was not conducted at 6 months given the absence of any case with ELC < 70 at that age.

3.2. Phase II analyses

Subsequent analyses focused on the subset of participants who had completed their 3-year diagnostic assessment at the time of data analysis (n = 424; 310 HR, 115 LR). One LR infant, who fell into trajectory 2, received a diagnosis of ASD at age 3. Given that he had no known family members with ASD, and non-familial cases may follow different trajectories than the ASD-HR group, his data were removed from subsequent analyses.

3.2.1. Diagnostic classification

The 3-year blinded assessment (M age = 38.64 months; SD = 3.11) yielded three primary diagnostic groups: (1) HR infants with ASD (ASD-HR: n = 77; 52 male); (2) HR infants without ASD (non-ASD-HR: n = 233; 116 male); and (3) non-ASD LR infants ('LR': n = 114; 59 male).

3.2.2. Socioeconomic status (SES)

Mean family SES (Hollingshead, 1975) differed significantly between HR(M = 48.01, SD = 12.67) and LR groups (M = 53.52, SD = 9.39), F(1,582) = 26.98, p < .001, although both were in the same stratum. Within the HR group, ASD (m = 47.02, SD = 13.68) vs. non-ASD (M = 48.53, SD = 13.51) subgroups did not differ, p = .54.

3.2.3. Sex distribution

There was no significant difference in sex distribution (m:f) for the combined sample (HR= 168:142 vs. LR= 59:55), X2 = 1.99, p = .66. However, within the HR group, sex ratio did differ between those with ASD (52:25) versus non-ASD (116:117) outcomes, x2 = 7.34; p < .01.

3.2.4. MSEL-ELC by diagnosis

Mean ages and MSEL-ELC scores at each assessment time-point are presented in Table 1. Significant group differences in MSEL-ELC scores emerged at each time-point; p's < .01. Post-hoc contrasts revealed differences between each group (ASD-HR < non-ASD HR < LR) at 12, and 24 months, and 3 years, but not at 6 months. No group differences for assessment age emerged; all p's > .14. None of these findings changed when family SES was entered as a covariate.

3.2.5. Prediction of 3-year diagnosis by MSEL-ELC trajectory

The association between cognitive trajectory and diagnosis was significant, x2 = 99.40; p < .001 (see Table 2). The majority (84%) of ASD-HR cases fell into Stable-Average or Declining trajectories, whereas most non-ASD-HR infants (92%) fell into Inclining or Stable-Average trajectories, with relatively few in the Declining trajectory. The majority of LR infants (80%) were in the Inclining trajectory, with only one Declining. Within the Declining trajectory, over half (58%) had ASD; the remainder were mostly non-ASD-HR (40%).

Table 2

Trajectory membership by diagnostic (Dx) group.

Trajectory group ASD-HR (n = 77) Non-ASD-HR (n = 233) LR (n =114) Total

Inclining (n) 12 121 90 223

% Dx group 15.6% 51.9% 78.9%

(% Trajectory group) (5.4%) (54.3%) (40.4%)

Stable-average (n) 39 94 23 156

% Dx group 50.6% 40.3% 20.2%

(% Trajectory group) (25.0%) (60.3%) (14.7%)

Declining (n) 26 18 1 45

% Dx group 33.8% 7.7% .9%

(% Trajectory group) (57.8%) (40.0%) (2.2%)

Note. ASD-HR: high-risk infants with ASD diagnosis at age 3; non-ASD-HR: high-risk infants with no diagnosis of ASD; LR: low-risk infants with no diagnosis of ASD.

3.2.6. Plateau or loss in MSEL raw scores

Declining standard scores may reflect developmental slowing, relative to developmental norms, rather than absolute skill loss. To explore the possibility of actual skill loss or developmental plateau, and its relation to diagnosis, we reviewed raw MSEL data in cases with available 3-year diagnostic outcome data as well as MSEL data at >2 time-points (n = 389; 71 ASD-HR, 211 non-ASD-HR, 107 non-ASD-LR). "Loss/Plateau" was defined as reduction in any domain raw score by >1 point, or no increase between consecutive assessments. Twenty-nine cases (7.5% of the available sample) met these criteria; 26 were from the HR group (9.2% of HR infants with available data); see Fig. 2. Significant associations emerged between loss/plateau and both group membership (HR vs. LR; x2 = 4.63; p < .05) and 3-year diagnosis (x2 = 24.13; p < .001 for 3 groups; x2 = 16.07; p < .001 for ASD-HR vs. non-ASD-HR). Although the 26 HR infants with loss/plateau were almost equally distributed between non-ASD (n = 11; 42.3%) and ASD (n = 15; 57.7%) subgroups, this represented only 5.2% of non-ASD-HR infants, versus 21.1% of the ASD-HR group. Loss/plateau was also highly associated with trajectory membership (Fisher's Exact = 30.97; p < .001), with 48.3% from the Declining trajectory and only 24.1% and 27.6% from the Inclining and Stable-Average trajectories, respectively. Loss/plateau occurred most often, but not exclusively, in language domains (15 instances each in RLand EL vs. 5 instances in each of FM and VR domains). Some, but not all cases with raw score loss/plateau were in the Declining trajectory (n = 14/41; 34.1% vs. only 3.5% from the Inclining and 5.4% from the Stable-Average trajectory). Of these 14 cases, 8 were ASD-HR (all met clinical criteria for ''Autistic Disorder,'' specifically) and 6 were non-ASD-HR. Two cases (both ASD-HR boys) had loss/plateau in all four MSEL domains, 5 (3 ASD-HR, 2 non-ASD-HR) had two affected domains, and the remainder had a single affected domain. When multiple domains were impacted, the timing of loss/plateau was consistent across domains; however, across participants the timing varied considerably (starting as early as 6-12 months), with a modal age of loss/plateau between 24 and 36 months. Four of the six non-ASD HR cases experienced loss/plateau in at least one language domain (2 RL, 2 EL, one of whom also had the FM domain affected), and 1 in both RL and EL; the other two had only the VR domain affected. The majority of these cases (n = 4) had a plateau rather than a loss.

3.2.7. Validity and distribution of ELC scores by diagnostic group

As in analytic phase I, we examined the presence of inter-domain scatter (i.e., domain t-scores > 2 SD apart) at each time point. For all time-points, Chi-square analyses revealed no association between diagnostic group membership and frequency of cases with inter-domain scatter (x2 = 1.13,2.71,4.30, and 1.80 at 6,12, and 24 months, and 3 years, respectively, p's range:

Fig. 2. Breakdown of cases with raw score "loss/plateau".

.12-.58). The proportion of cases with ELC <70 for each diagnostic group at 6,12,24 months, and 3 years was as follows: HR-ASD: 0%, 3.1%, 22.7%, and 26.4% of cases, respectively; non-ASD-HR: 0%, 1.1%, 4.1%, and 3.5%, respectively; LR: 0 cases at all age points. At both 24 months and 3 years, significant associations emerged between proportion of cases with ELC < 70 and diagnostic group (Fisher's Exact = 31.28, and 43.62, respectively, both p's < .001). The limited number of cases with ELC < 70 precluded analyses at 6 (n = 0) and 12 (n = 4) months.

4. Discussion

We examined developmental trajectories of infants at high and low risk for ASD and assessed whether variation in trajectory membership was associated with 3-year diagnosis. Using overall MSEL standard scores, we identified three trajectories from 6 months to 3 years: (1) 'Inclining', (2) 'Stable-Average', and (3) 'Declining'. These parallel three of the four trajectories (accelerated, normative, language/motor delayed, and declining) reported by Landa et al. (2012). Contrary to Landa and colleagues, however, we did not identify distinct sub-groups with delayed versus declining trajectories. This discrepancy may be explained by the use of different statistical methods, our use of a larger combined HR and LR sample, or by a cohort effect. It also remains possible that the delayed and declining trajectories described by Landa and colleagues were represented by a single declining group in our study, more consistent with their 3-class model.

A variety of developmental patterns emerged in our sample; three distinct trajectories significantly discriminated between groups based on 3-year diagnosis. One-third of our ASD group fell into the Declining trajectory, comparable to the 41% reported by Landa and colleagues (2012). Consistent with the most recent estimates of rates of intellectual disability in the general ASD population (i.e., 31% as reported by the CDC, 2014) almost one-third (26%) of our HR participants with ASD had a standard score of 70 or less on the MSEL-ELC at age 3. Note, however, that two-thirds of those with ASD exhibited age-appropriate developmental levels in the first three years of life. Indeed, although consistent with findings from other studies of high-risk sibling samples (Landa et al., 2012), our ASD group demonstrated somewhat higher developmental functioning at age 3 than ASD samples obtained through other sources such as clinical referral (Turner & Stone, 2007) or community screening (Ventola et al., 2006), which may reflect differences in ascertainment. The vast majority of non-ASD HR infants also had intact development, but a small number were represented by a declining trajectory, demonstrating heterogeneity in developmental progression among HR infants, irrespective of ASD or non-ASD outcomes.

A pattern of performance declining from average to well below age expectations by age two was highly predictive of a diagnosis of ASD: more than half of the cases with a declining trajectory received an ASD diagnosis. However, over one-third of the declining group consisted of HR infants without an ASD diagnosis, indicating that this pattern is not unique to those with ASD outcomes. A pattern of declining cognitive development may indicate a broader spectrum of impairment, at least within this high-risk group.

Our findings highlight the importance of longitudinal research into the early development of non-ASD siblings, particularly in light of accumulating evidence of sub-clinical ASD-related symptoms, including social-communication challenges and rigid behavioral styles, often referred to as the broader autism phenotype ('BAP'; Ben-Yizhak et al., 2011; Bishop et al., 2006; Georgiades et al., 2013; Messinger et al., 2013), but also with reference to other developmental and/or learning difficulties. Only one-third of the non-ASD-HR infants with declining trajectories plus raw loss/plateau in the current study were identified at 12 months as falling into the 'BAP' group in our recent cluster analysis (Georgiades et al., 2013), highlighting the complexity of the developmental profiles in these high-risk children. None of our participants was formally assessed for other disorders that would affect test performance either directly (e.g., learning disability, language impairment), or indirectly (e.g., ADHD, anxiety). As we follow this sample, it remains possible that non-ASD siblings with developmental slowing will manifest such difficulties. Even diagnoses of higher functioning ASD or Social-Communication Disorder (APA, 2013; Happe, 2011) remain possible outcomes.

We emphasize the distinction between a decline in relative standing over time (i.e., declining standard scores) and an actual loss of skill or failure to progress (i.e., flat or declining raw scores), which may be more indicative of developmental regression. Within our Declining trajectory, just over one-third experienced any raw score loss/plateau. Although almost exclusive to HR infants, this was not specific to those with an ASD outcome, at least not by age three. Note, however, that the nature of loss/plateau was narrowly defined in our study, and varied considerably across cases, making it difficult to compare with reports using more conventional definitions of ''autistic regression'' that typically focus on language and social-communication (as outlined in the ADI-R; Lord et al., 1994). In our sample, loss/plateau varied by degree, timing, and domains affected, supporting the notion of regression as a variable phenomenon. While some instances of declining performance might be accounted for, at least in part, by increasing challenges with test-taking abilities (e.g., reduced attention/focus, increased behavioral dysregulation, emerging non-compliance; Akshoomoff, 2006) rather than loss of skills per se, such behavioral changes would nonetheless reflect increasing functional impairment, and presumably have negative effects on learning.

Consistent with our findings, developmental regression is typically reported to occur between 12 and 30 months, and specifically in cases of Autistic Disorder (vs. milder forms of ASD). Although skill loss is generally thought to be highly specific to ASD, this is based on studies that compare ASD cohorts to other clinically ascertained (e.g., language impairment) or typically developing groups not genetically at-risk for ASD (Lord, Shulman, & DiLavore, 2004; Pickles et al., 2009). Our rate of skill loss/plateau in approximately 20% of participants with ASD is within the range generally reported for autistic regression (Barger et al., 2013; Ekinci et al., 2012; Stefanatos, 2008; Yirmiya & Charman, 2010), but considerably lower than the 75%

based on social-communication skills (which did not correlate with MSEL performance) that was recently reported in younger siblings of children with ASD (Ozonoff et al., 2010).

Recognizing that our data may not reflect the pervasive regression that is sometimes associated with autism, we have identified developmental loss or plateau in a variety of domains, suggesting that the emergence of ASD can be associated with more than one type of atypical trajectory, and highlighting that ''regression'' is not a unitary construct. Declining developmental trajectories warrant careful clinical monitoring, with the following considerations: (1) a declining developmental trajectory or absolute skill loss in a younger sibling of a child with ASD is a risk factor for ASD, but can also be associated with other developmental concerns that warrant attention and possibly intervention; and (2) absence of a declining trajectory does not preclude a future diagnosis of ASD. Early signs of developmental slowing warrant referral for further assessment (Zwaigenbaum et al., 2009), and parents should be supported to incorporate techniques into their daily routines to enhance social-communication development, such as strategies from infant development programs, preschool speech-language services, or other ASD-focused approaches for toddlers that are emerging in the literature (Brian et al., 2012; Dawson et al., 2010; Siller et al., 2014). Targeted intervention at this very early age may attenuate sub-optimal outcomes in at least some high-risk children. Evidence of diverging developmental pathways as early as 12-24 months underscores that siblings of children with ASD are in particular need of careful monitoring early in life, and that an apparently typical profile at 6 months does not 'guarantee' later normative development (Bryson et al., 2007; Rozga et al., 2011; Zwaigenbaum et al., 2005).

We acknowledge several limitations to the current study and considerations for the interpretation of our findings. First, the validity of the MSEL-ELC as an index of overall cognitive development may be questioned, given the high probability of inter-domain scatter, particularly in our high-risk group. We explored this potential challenge by examining inter-domain scatter and found no significant association with group membership, suggesting that the ELC behaves similarly across groups. We chose to focus on the ELC in order to capture a broad, overall characterization of developmental functioning and progression in our high-risk group, but we recognize the importance of examining different domains separately. Further analyses are underway to examine developmental trajectories in each of the primary MSEL domains. We also acknowledge the unexpected finding of apparent gains in developmental standing in the LR controls from solidly within average limits at 6 months, to >1 SD above average by 24 months. One possibility for this is the somewhat weaker psychometric properties at 6 vs. 24 months in the standardization sample (i.e., lower split-half internal consistency (r = .91 vs. .94), and larger standard error of measurement (4.5 vs. 3.7; Mullen, 1995). Smaller group differences at 6 months may also be explained, at least partially, by the presence of floor effects in the high-risk sample. Specifically, the very small number of items needed to obtain an average score at 6 months, together with our a priori exclusion of infants with significant neurological and/or sensory impairments, resulted in none of our cases obtaining scores >2 SD below average at this age. The relatively low variability in scores at 6 vs. 24 months (as depicted in Table 1) would lend support to this explanation. While the MSEL is a widely used tool in the assessment of early development in ASD, and particularly favored in infant sibling research, it remains to be seen whether this measure accurately captures the nuances of development in this population. Future research may benefit from the use of other psychometrically sound measures of intellectual functioning (e.g., Bayley Scales of Infant and Toddler Development-III, Bayley, 2005; Stanford Binet-5, Roid, 2003; Merrill Palmer-Revised; Roid & Sampers, 2004) to ensure that the developmental patterns described to date in these high risk samples are more than artifacts associated with the psychometric properties of one tool.

Second, the representativeness of our sample may be limited by several potential issues: (1) higher mean SES in the LRthan HR group. This may be explained by the intense commitment required for longitudinal research participation that may have selected for a somewhat more advantaged control group, whereas parents in the more socioeconomically diverse HRgroup may have been motivated to participate in order to have their younger children monitored; (2) relatively high MSEL-ELC scores. Our LR and non-ASD-HR groups both had high mean MSEL-ELC scores relative to published norms. Plausible explanations for this include the possibility of selection bias of families interested in or able to participate in research, the fact that the MSEL was not standardized on Canadian children, or the Flynn Effect (i.e., systematic increases in population IQperformance over time; see Flynn, 2007) having contributed to elevated scores relative to published norms. (3) Finally, our enrollment based solely on familial ASD risk may limit the generalizability of our findings. It remains possible that the inclusion of other samples (i.e., clinically referred, simplex cases) may yield different results. This caveat is underscored by the high overall developmental functioning in our ASD sample relative to clinically ascertained or screen-positive cases described in the literature (but generally consistent with other HR sibling samples; e.g., Landa et al., 2012). Evidence is beginning to implicate different genetic mechanisms in the emergence of ASD in multiplex vs. simplex families (Zwaigenbaum et al., 2012), supporting the notion that phenotypic expression may also differ systematically across these groups.

Although our findings generate provocative questions about the nature of developmental progression and slowing in high risk siblings, our ability to draw conclusions about regression is limited by not having explored social and/or behavioral regression, and by our conservative definition of skill 'loss' (i.e., >1 point, whereas gains of 4-10 points are expected across the time-spans examined; Mullen, 1995). We may thus have underestimated the true prevalence of functional loss. Continued efforts toward the characterization of regression during the emergence of ASD, including careful delineation of specific features of skill loss (i.e., timing, domain(s), duration, etc.), may be informative for both monitoring and intervention purposes. We are following our non-ASD-HR sample, with particular attention to those with declining developmental trajectories, in order to determine whether this pattern is indeed predictive of other developmental, learning, or mental health disorders in the school-age years.

References

Akshoomoff, N. (2006). Use of Mullen scales of early learning for the assessment of young children with ASD. Special Issue: ASD. Child Neurology, 12, 269-277.

American Psychiatric Association. (2013). Diagnostic and statistical manual of mental disorders (5th ed.). Washington, DC: Author.

American Psychiatric Association. (2000). Diagnostic and statistical manual of mental disorders (4th ed text rev; DSM-IV-TR). Washington, DC: Author.

Baird, G., Douglas, H., & Murphy, M. S. (2011). Recognising and diagnosing autism in children and young people: summary of NICE guidance. British Medical Journal, 343, d6360.

Barger, B. D., Campbell, J. M., & McDonough, J. D. (2013). Prevalence and onset of regression within ASD: A meta-analytic review. Journal of Autism Developmental Disorders, 43, 817-828.

Bayley, N. (2005). Bayley scales of infant and toddler development (third ed.). San Antonio, TX: Pearson.

Ben-Yizhak, N., Yirmiya, N., Seidman, I., Alon, R., Lord, C., & Sigman, M. (2011). Pragmatic language and school related linguistic abilities in siblings of children with autism. Journal of Autism and Developmental Disorders, 41, 750-760.

Bishop, D. M. V., Mayberry, M., Wong, D., Maley, A., & Hallmayer, J. (2006). Characteristics of the broader phenotype in autism: A study of siblings using the children's communication checklist-2. American Journal of Medical Genetics, Part B: Neuropsychiatric Genetics, M1B(2), 117-122.

Brian, J., Smith, I. M., McCormick, T., Dowds, E., Longard, J., Dhooge, S., et al. (2012, May). The Social ABC's for toddlers with suspected autism: pilot evaluation of a parent-mediated intervention. Poster presented at the International Meeting For Autism Research, Toronto, ON.

Bryson, S. E., Zwaigenbaum, L., Brian, J. A., Roberts, W., Szatmari, P., Rombough, V., et al. (2007). A prospective case series of high-risk infants who developed autism. Journal of Autism and Developmental Disorders, 37(1), 12-24.

Centers for Disease Control and Prevention. (2014). Prevalence of autism spectrum disorder among children aged 8 years - Autism and Developmental Disabilities Monitoring Network, 11 sites, United States, 2010. Morbidity and Mortality Weekly Report, Surveillance Summary, 63(2), 1-20.

Chakrabarti, S., & Fombonne, E. (2005). Pervasive developmental disorders in preschool children: confirmation of high prevalence. American Journal of Psychiatry, 162,1133-1141.

Dawson, G., Rogers, S., Munson, J., Smith, M., Winter, J., Greenson, J., et al. (2010). Randomized, controlled trial of an intervention for toddlers with autism: The Early Start Denver Model. Pediatrics, 125(1), e17-e23.

Drumm, E., & Brian, J. A. (2013). The developing language abilities and increased risks of 'unaffected' siblings of children with autism spectrum disorder. Neuropsychiatry, 3(5), 513-524. http://dx.doi.org/10.2217/npy.13.65

Ekinci, O., Arman, A. R., Melek, I., Bez, Y., & Berkem, M. (2012). The phenomenology of autistic regression: Subtypes and associated factors. European Child and Adolescent Psychiatry, 21, 23-29.

Flynn, J. R. (2007). What is intelligence? Beyond the Flynn effect Cambridge. UK Cambridge University Press.

Georgiades, S., Szatmari, P., Zwaigenbaum, L., Bryson, S. E., Brian, J., Roberts, W., et al. (2013). A prospective study of autistic-like traits in unaffected siblings of probands with ASD. Journal of the American Medical Association Psychiatry, 70, 42-48.

Happe, F. (2011). Criteria, categories, and continua: autism and related disorders in DSM-5. Journal of the American Academy of Child and Adolescent Psychiatry, 50(6), 540-542.

Harris, S., & Handleman, J. S. (2000). Age and IQat intake as predictors of placement for young children with autism: a four- to six-year follow-up. Journal of Autism and Developmental Disorders, 30(2), 137-142.

Hollingshead, A. A. (1975). Four-factor index of social status. New Haven, CT: Yale University (unpublished manuscript).

Howlin, P., Goode, S., Hutton, J., & Rutter, M. (2000). Adult outcomes for children with autism. Journal of Child Psychology and Psychiatry, 45(2), 212-229.

Ingesson, S. G. (2006). Stability of IQ measures in teenagers and young adults with developmental dyslexia. Dyslexia, 12(2), 81-95.

Kelley, E., Naigles, L., & Fein, D. (2010). An in-depth examination of optimal outcome children with a history of ASD. Research in Autism Spectrum Disorders, 4(3), 526-538.

Landa, R. J., Gross, A. L., Stuart, E. A., & Bauman, M. (2012). Latent class analysis of early developmental trajectory in baby siblings of children with autism. Journal of Child Psychology and Psychiatry, 53, 986-996.

Landa, R J., Gross, A. L., Stuart, E. A., & Faherty, A. (2013). Developmental trajectories in children with and without ASD: The first 3 years. Child Development, 84, 429-442.

Landa, R J., Holman, K. C., & Garrett-Mayer, E. (2007). Social and communication development in toddlers with early and later diagnosis of ASD. Archives of General Psychiatry, 64, 853-864.

Lord, C., Risi, S., DiLavore, P., Shulman, C., Thurm, A., & Pickles, A. (2006). Autism from 2 to 9 years of age. Archives of General Psychiatry, 63, 694-701.

Lord, C., Risi, S., Lambrecht, L., Cook, E. H., Leventhal, B. L., DiLavore, P. C., et al. (2000). The autism diagnostic observation schedule-generic: A standard measure of social and communicative deficits associated with the spectrum of autism. Journal of Autism and Developmental Disorders, 30, 205-223.

Lord, C., Rutter, M., DiLavore, P., & Risi, S. (1999). Autism diagnostic observation schedule: Manual. Los Angeles: Western Psychological Services.

Lord, C., Rutter, M., & LeCouteur, A. J. (1994). Autism diagnostic interview-revised: a revised version of a diagnostic interview for caregivers of individuals with possible pervasive developmental disorders. Journal of Autism and Developmental Disorders, 24, 659-685.

Lord, C., & Schopler, E. (1989a). Stability of assessment results of autistic and nonautistic language-impaired children from preschool years to early school age. Journal of Child Psychology and Psychiatry, 30, 575-590.

Lord, C., & Schopler, E. (1989b). The role of age at assessment, developmental level, and test in the stability of intelligence scores in young autistic children. Journal of Autism and Developmental Disorders, 19(4), 483-499.

Lord, C., Shulman, C., & DiLavore, P. (2004). Regression and word loss in ASD. Journal of Child Psychology and Psychiatry, 45(5), 936-955.

Messinger, D., Young, G. S., Ozonoff, S., Dobkins, K., Carter, A., Zwaigenbaum, L., et al. (2013). Beyond autism: a baby siblings research consortium study of high-risk children at three years of age. Journal of the American Academy of Child and Adolescent Psychiatry, 52(3), 300-308.

Mortensen, E. L., Andressen, J., Krusse, E., & Sanders, S. A. (2003). IQ stability: The relation between child and young adult intelligence test scores in low-birthweight samples. Scandinavian Journal of Psychology, 44(4), 395-398.

Mullen, E. M. (1995). Mullen scales of early learning (AGS Edition). Circle Pines, MN: American Guidance.

Nagin, D. S. (1999). Analyzing developmental trajectories: a semi-parametric, group based approach. Psychological Methods, 4,139-177.

Ozonoff, S., Iosif, A. M., Baguio, F., Cook, I. C., Moore, M., Hutman,T., etal. (2010). A prospective study of the emergence of early behavioral signs of autism. Journal of the American Academy of Child and Adolescent Psychiatry, 49(3), 256-266.

Ozonoff, S., Young, G. S., Carter, A., Messinger, D., Yirmiya, N., & Zwaigenbaum et al. (2011). Recurrence risk for autism spectrum disorders: a Baby Siblings Research Consortium study. Pediatrics, 128, e488-e495.

Pickles, A., Simonoff, E., Conti-Ramsden, G., Falcaro, M., Simkin, Z., Charman, T., et al. (2009). Loss of language in early development of autism and specific language impairment. Journal of Child Psychology and Psychiatry, 50(7), 843-852.

Ramsden, S., Richardson, F. M., Josse, G., Thomas, M., Ellis, C., Shakeshaft, C., et al. (2011). Verbal and non-verbal intelligence changes in the teenage brain. Nature, 479,113-116.

Risi, S., Lord, C., Gotham, K., Corsello, C., Chrysler, C., Szatmari, P., et al. (2006). Combining information from multiple sources in the diagnosis of ASD. Journal of the American Academy of Child and Adolescent Psychiatry, 45,1094-1103.

Roid, G. H. (2003). Stanford-Binet intelligence scales (Fifth Edition). Itasca, IL: Riverside Publishing.

Roid, G., & Sampers, J. (2004). Merrill-Palmer-revised scales of development. Wood Dale, IL: Stoelting Co.

Rozga, A., Hutman, T., Young, G. S., Rogers, S., Ozonoff, S., Dapretto, M., et al. (2011). Behavioral profiles of affected and unaffected siblings of children with autism: contribution of measures of mother-infant interaction and nonverbal communication. Journal of Autism and Developmental Disorders, 41(3), 287-301.

Siller, M., Morgan, L., Turner-Brown, L., Baggett, K. M., Baranek, G. T., Brian, J. A., et al. (2014, July). Designing studies to evaluate parent-mediated interventions for toddlers with autism spectrum disorder. Journal ofEarly Intervention. http://dx.doi.org/10.1177/1053815114542507

Stefanatos, G. A. (2008). Regression in ASD. Neuropsychological Review, 18, 305-319.

Sutera, S., Pandey, J., Esser, E., Rosenthal, M. A., Wilson, L. B., Barton, M., et al. (2007). Predictors of optimal outcome in toddlers diagnosed with ASD. Journal of Autism and Developmental Disorders, 37(1), 98-107.

Turner, L. M., & Stone, W. L. (2007). Variability in outcome for children with an ASD diagnosis at age 2. Journal of Child Psychology and Psychiatry, 48(8), 793-802.

Ventola, P. E., Kleinman, J., Pandey, J., Barton, M., Allen, S., Green, J., et al. (2006). Agreement among four diagnostic instruments for ASD in toddlers. Journal of Autism and Developmental Disorders, 36(7), 839-847.

Whitaker, S., & Taylor, S.J. (2008). The stability of IQin people with low intellectual ability: An analysis of the literature. Intellectual and Developmental Disability, 46(2), 120-128.

Yirmiya, N., & Charman, T. (2010). The prodrome of autism: early behavioral and biological signs, regression, peri- and post-natal development and genetics. Journal of Child Psychology and Psychiatry, 51(4), 432-458.

Zwaigenbaum, L., Bryson, S., Lord, C., Rogers, S., Carter, A., Carver, L., et al. (2009). Clinical assessment and management of toddlers with suspected ASD: insights from studies of high-risk infants. Pediatrics, 123(5), 1383-1391.

Zwaigenbaum, L., Bryson, S. E., Rogers, T., Roberts, W., Brian, J., & Szatmari, P. (2005). Behavioural manifestations of autism in the first year of life. International Journal of Developmental Neuroscience, 23, 143-152.

Zwaigenbaum, L., Bryson, S. E., Szatmari, P., Brian, J., Smith, I. M., Roberts, W., et al. (2012). Sex differences in children with ASD identified within a high-risk infant cohort. Journal of Autism and Developmental Disorders, 42(12), 2585-2596.