Medscape Medical News
Norra MacReady
May 05, 2014
In an analysis of more than 2 million families, genetics appeared to contribute 50% of the risk for an autism spectrum disorder (ASD), with environmental influences accounting for the remaining 50%, according to a report in an article publishedin the May 7 issue of JAMA.
ASD risk increased as genetic relatedness within families increased. Full siblings had the highest risk, followed by maternal half siblings, paternal half siblings, and cousins.
“[T]his is, to our knowledge, the largest population-based longitudinal study evaluating familial risk of ASD,” write Sven Sandin, MSc, from the Department of Medical Epidemiology and Biostatistics, Karolinska Institutet, Stockholm, Sweden, and colleagues.
The researchers studied the records of all children born in Sweden between January 1, 1982, and December 31, 2006. They determined family relationships from the Swedish Multigeneration Register, which identifies the parents of all children born in Sweden from 1932 on. At 4 years of age, all Swedish children undergo a mandatory developmental assessment, and any child with a suspected developmental disorder is referred to specialists for further examination. Diagnostic information is entered into the National Patient Register and becomes a matter of public record.
The authors used these data to determine the family clustering of ASD in all twins, full and half siblings, and cousins born in Sweden between 1982 and 2006. The Swedish Multigeneration Register defined cousins as children with grandparents but no parents in common. Only children were excluded from the analysis.
Overall, the researchers included 2,049,973 children in the study who were born between 1982 and 2006. There were 2,642,064 full-sibling pairs, 432,281 maternal half-sibling pairs, 445,531 paternal half-sibling pairs, 37,570 twin pairs, and 5,799,875 cousin pairs.
Of the children born during the study period, 14,516 had ASD, including 5689 (39%) with autistic disorder. The male-to-female ratios for all ASD cases and autistic disorder were 2.7 and 2.4, respectively.
The table shows the cumulative probabilities and RRR of an ASD diagnosis by 20 years of age, according to family relationship. The authors defined relative recurrence risk (RRR) as the risk for ASD in a child with a sibling with ASD (exposed) relative to that of a child with no siblings with ASD (unexposed). It was calculated using a Cox proportional hazards regression scale, with sibling-attained age as the underlying time scale, adjusted for confounding variables, including parental psychiatric history and age at birth and the child’s birth year and sex. They also used the Cox model to calculate the cumulative probability, or prevalence, of ASD up to 20 years of age.
Table. Cumulative Probability and RRR of an ASD Diagnosis Based on Having a Relative With ASD
| Relationship | Cumulative Probability | RRR (95% Confidence Interval [CI])* |
| Monozygotic twins | 59.2% | 153.0 (56.7 – 412.8) |
| Dizygotic twins | 12.9% | 8.2 (3.7 – 18.1) |
| Full sibling | 12.9% | 10.3 (9.4 – 11.3) |
| Maternal half-sibling | 8.6% | 3.3 (2.6 – 4.2) |
| Paternal half-sibling | 6.8% | 2.9 (2.2 – 3.7) |
| Cousins | 2.6% | 2.0 (1.8 – 2.2) |
| No sibling with ASD | 1.2% |
*Per 100,000 person-years. Adjusted for 5-year birth cohorts, sex, parental age, and parental psychiatric history.
The researchers found no statistically significant difference in RRR between boys and girls.
In calculating heritability, the researchers determined a model that included additive genetic and nonshared environmental parameters to be the best fit of the various submodels used, such as those that excluded the dominant genetic term or the additive genetic and nonshared environment. With this model, heritability for ASD was estimated at 0.50 (95% CI, 0.45 – 0.56). Nonshared environmental influence also was estimated at 0.50 (95% CI, 0.44 – 0.55). For autistic disorder, the same model yielded a heritability estimate of 0.54 (95% CI, 0.44 – 0.64).
These findings suggest that “genetic factors explain half of the risk for autism,” the authors write. “This is considerably lower than the 90% in earlier twin studies, and closer to the 38% (95% CI, 14%-67%) reported in a recent California twin study, but estimated with substantially higher precision.” They attributed these discrepancies to differences in sampling, case ascertainment, and analytical approach among the different studies.
Limitations of this study include the lack of information on parental educational and socioeconomic status and the inability to compare twins raised to together and apart, “which could have contributed information to the estimation of the shared and nonshared environment.”
This research represents “the first attempt to provide both ASD recurrence and heritability estimates from the same population,” write Diana E. Schendel, PhD, from the Section for Epidemiology, Department of Public Health, Aarhus University, Denmark, and colleagues in an accompanying editorial . It is also the first attempt to go back a generation and include siblings and cousins in the estimates of ASD heritability. The real novelty of the study “is in the delivery of a single, population-based ‘package’ of results of extended familial risk based on a very large sample.”
Parents of one child with ASD can be reassured from these findings that the risk for recurrence in another child is lower than suggested by some earlier studies, the editorialists write. They conclude that although many unanswered questions remain, this work “supports appreciation of the importance of genetic factors in ASD and adds substantial impetus to the growing attention to environmental influences in ASD etiology.”
The study authors and editorialists have disclosed no relevant financial relationships.
JAMA. 2014;311:117-1777; 1738-1739. Article abstract, Editorial extract