Twin and Trio Analyses Separate Direct From Indirect Genetic Effects in ADHD and ASD Traits
Polygenic prediction of ADHD and ASD traits has long raised a basic interpretive problem: do the scores capture a child’s own inherited liability, or do they also reflect the genetic architecture of the family environment shaping development? A new triangulated analysis in the Twins Early Development Study tackles that question across childhood through young adulthood, and across parent, teacher, and self-reported traits, using two complementary designs that point in the same direction more often than not.
The study combined twin-comparison methods with trio-based polygenic score analyses in up to 7,259 offspring with observed or imputed parental genetic data. Eight polygenic scores were tested against repeated measures of attention-deficit/hyperactivity disorder and autism spectrum disorder traits from age 4 to 26. The design matters because it lets investigators ask a sharper question than standard association studies can answer: whether the score is acting through the child’s own genome, through parental genotypes that shape the rearing environment, or through both.
In practice, the answer was clearest for within-trait prediction. Both analytical strategies consistently supported direct genetic effects of ADHD and ASD polygenic scores on the matching traits. Evidence for indirect genetic effects within the same trait was limited. In a field where inherited risk and family-level transmission often blur together, that separation is clinically meaningful. It suggests that much of the polygenic signal linking ADHD scores to ADHD traits, and ASD scores to ASD traits, is rooted in the child’s own genetic liability rather than in parental genetic influences operating through the home environment alone.
Developmental Follow-Up Reinforces a Stable Signal Across Childhood and Young Adulthood
The repeated-measures design adds another layer of value. Traits were assessed from age 4 through 26, allowing the investigators to ask whether the relative contribution of direct and indirect genetic effects changed as children matured into adults. The pattern remained consistent across development: direct effects were repeatedly supported, while within-trait indirect effects stayed weak. In a developmental context, that stability argues against the idea that family-mediated polygenic influence dominates one stage and disappears at another.
A single sentence captures the core result.
Within-trait polygenic prediction of ADHD and ASD traits was driven mainly by the child’s own inherited genetic liability.
That finding has practical implications for how researchers interpret polygenic scores in neurodevelopmental phenotypes. If a score mainly reflects direct effects, then models that treat the score as a child-level risk marker are closer to biological reality than models that fold the association into broad familial transmission. The distinction is especially important when investigators use these scores in mediation analyses, etiologic studies, or attempts to explain trait continuity over time.
Cross-Trait Signals Point to Family-Mediated Effects, Especially in Parent Reports
The picture changed when the analysis moved beyond same-trait prediction. Cross-trait indirect effects emerged for several polygenic scores, and these signals were particularly apparent in parent reports. Some of these effects were unexpectedly negative. In other words, the parental genetic background associated with one neurodevelopmental trait sometimes tracked lower levels of a different reported trait rather than higher levels. Such findings complicate any assumption that family genetic influences move in a single, uniformly risk-amplifying direction.
Those cross-trait patterns are where indirect genetic effects become more plausible. Parental genotypes can shape the environment in ways that alter how offspring traits are expressed, noticed, or reported. That may include the structure of the household, patterns of interaction, parental expectations, or the kinds of behaviors that are more likely to be observed and labeled by adults. Parent reports are therefore especially vulnerable to family-level genetic confounding, not because they are unreliable, but because they sit closer to the interface between observable behavior and the environment in which that behavior is interpreted.
Teacher and self-reports offer a different vantage point, and the study’s multi-reporter design helps expose where family-mediated influences are strongest. The repeated appearance of indirect effects in parent reports, alongside more limited within-trait indirect evidence overall, suggests that the reporter is not a neutral technical detail. It can change the apparent genetic architecture of the phenotype.
Why the Trio Design Matters for Neurodevelopmental Genetics
Trio-based analyses are valuable because they use parental genotypes to distinguish transmitted from non-transmitted genetic influence. Twin-comparison methods, by contrast, leverage the differing genetic sharing within twin pairs to infer whether a measured association is more compatible with direct inheritance or broader familial pathways. Bringing both approaches together gives a more disciplined view of causality than either method alone.
For ADHD and ASD traits, that triangulation converged on a consistent message: the within-trait polygenic associations are mostly direct. Cross-trait associations are more likely to include some indirect genetic component. The result does not eliminate family-level genetic influence from neurodevelopmental research. It narrows where that influence appears to matter most.
A second sentence stands alone because the data demand it.
Indirect genetic effects were not the main explanation for same-trait polygenic prediction.
This matters for future work on developmental risk stratification, gene-environment interplay, and the interpretation of polygenic scores in longitudinal cohorts. Researchers often treat a polygenic score as if it were a simple personal liability index. These findings show a more specific reality: for ADHD and ASD traits, that simplification is defensible for within-trait prediction, but not for every cross-trait relationship or every reporter.
Limits of the Signal and the Remaining Measurement Problem
The findings are careful rather than sweeping. Evidence for within-trait indirect effects was limited, not absent in every possible comparison, and the cross-trait signals varied by score and reporter. The study also depended on observed or imputed parental genetic data, which is useful but not identical to a fully observed trio sample. Repeated measurement strengthens inference, yet reporter effects and phenotype definition still shape what the scores appear to predict.
Even so, the study pushes the field toward a more precise framework for interpreting polygenic prediction in neurodevelopmental traits. The core lesson is not that indirect effects do not exist, but that they are not the dominant explanation for matching ADHD and ASD score-trait associations. For clinicians and researchers reading genetic association studies, that distinction changes how much weight to place on family-mediated pathways versus the child’s inherited biology.
The next challenge is not whether polygenic scores can predict neurodevelopmental traits. It is how much of that prediction survives when the analysis is forced to separate the child’s genome from the genetic environment created by the parents, one reporter, and one developmental stage at a time.