Birth Weight in ASD-Affected Twin Pairs

Background:

Multiple studies have associated low or very low birth weight with Autism Spectrum Disorders (ASDs). If low birth weight is connected with ASDs, it is reasonable to hypothesize that, in ASD-discordant twin pairs, the twin with the smaller birth weight will more often be the affected twin, and that ASD-concordant twin pairs will have lower average birth weights than ASD-discordant pairs. Additionally, in ASD-concordant pairs, the twin with the smaller birth weight may be more severely affected. 

Objectives:

To examine the relationship of relative birth weight in ASD-affected twin pairs and ASD severity in a large twin sample.  

Methods:

Twin pairs were identified through the California Department of Development Services in cooperation with the California Department of Public Health (CDPH) and were eligible if at least one twin met ASD criteria without an associated neurogenetic condition. Assessments included the ADI-R, ADOS, IQ testing, Social Responsiveness Scale (SRS), Peabody Picture Vocabulary Test (PPVT), and Vineland Adaptive Behavior Scale (VABS). Birth weight was extracted from CDPH birth records.

Only same-sex pairs were included in this analysis. Univariate analyses of covariance (ANCOVAs) controlling for gestational age were used to compare birth weights in affected and unaffected twins, as well as within pair birth weight averages. Paired sample t-tests compared birth weights in affected and unaffected co-twins. To examine birth weight in relation to measures of severity, partial correlations controlling for gestational and gender were used. 

Results:

Analyses included 122 same-sex pairs (N=72 ASD-discordant pairs and N=50 ASD-concordant pairs).

In ASD-discordant pairs, the affected twin had a lower birth weight in only 7/16 MZM pairs (43.8%), 2/5 MZF pairs (40%), 19/42 DZM pairs (45.2%), and 3/9 DZF pairs (33.3%). Comparing mean birth weights of unaffected twins to affected twins, ANCOVAs controlling for gestational age showed no significant differences in either males or females (mean=2489 grams, N=142 affected males vs. mean=2557 grams, N=57 unaffected males; and mean=2403 grams, N=28 affected females vs. mean=2355 grams, N=14 unaffected females). Paired sample t-tests showed no significant differences in birth weights between affected and unaffected twins for DZ, ASD-discordant twin pairs (N=51) or for MZ, ASD-discordant twin pairs (N=21).

Overall, ANCOVA controlling for gestational age showed no significant difference in the within pair birth weight averages for ASD-concordant (mean=2458 grams, N=49) compared to ASD-discordant pairs (mean=2508 grams, N=71). ANCOVA remained non-significant for male only and female only comparisons.

Controlling for gestational age and gender, birth weight did not correlate with ASD severity in affected individuals as measured by ADOS severity, IQ, SRS, PPVT, or VABS. Furthermore, ANCOVA controlling for gestational age and gender demonstrated no significant difference in birth weight comparing those diagnosed with strict autism (N=81) versus autism spectrum disorder (N=17).    

 Conclusions:

Results suggest relative birth weights do not distinguish twins with ASDs from their unaffected co-twins in ASD-discordant pairs. Within pair averages of birth weights were not significantly different in ASD-concordant compared to ASD-discordant twin pairs. Additionally, birth weight does not correlate with ASD severity in this cohort of twins.