In this paper, we examined gene × parenting interactions as predictors of the development of children’s noncompliance and aggression in early childhood. Specifically, we tested whether two genetic variants on the serotonin transporter gene (SLC6A4)—LPR and STin2—would moderate the effect of parenting on initial levels and rates of change in children’s noncompliance and aggression, as rated by non-parental caregivers in early childhood.

In a sample of 138 children, maternal parenting was observed during parent-child interactions during an 18-month laboratory visit. Noncompliance and aggression was measured using caregivers’ reports at 18, 30, 42, and 54 months of age.

We used mixed models to estimate children’s trajectories of noncompliance and aggression, and found that although both outcomes were moderately to highly correlated at each assessment, aggression decreased across early childhood whereas noncompliance showed a quadratic trajectory, increasing at the start of the study and peaking at 42 months.

We hypothesized that the haplotypes that combine information from LPR and STin2 (similar to a gene × gene interaction) would provide better prediction than either variant alone. Consequently, we tested moderation by genetics for each variant individually, as well as combined as haplotypes, which are genetic variants that tend to be inherited together.

We found that LPR moderated the effects of parenting on noncompliance at the start of the study, but that this difference diminished over time. Relative to children who were homozygous for the “long” (L) LPR allele, children with at least one copy of the “short” (S) LPR allele were more susceptible to the effects of parenting on noncompliance— but not aggression— at 18 months. STin2 did not predict either outcome.

Then we predicted noncompliance using haplotypes that combined information from LPR (short vs. long) and STin2 (10 vs. 12 repeat). Based on research suggesting that STin2 moderated gene expression for individuals carrying the “S” variant of LPR, we separated the LPR “S” carriers into two groups: individuals with the 10 repeat variant of STin2, and those with the 12 repeat variant of STin2; the LL group was not subdivided in this way. These three haplotype groups provided more differentiated prediction relative to the LPR variant alone. Although the S10 group was not very sensitive to parenting at the start of the study, parenting became more influential for this group over time. In contrast, the effect of parenting on the S12 group was high (relative to the LL group) across the entire study period.

The findings from this study provide evidence that the use of haplotypes can provide better prediction relative to individual variants alone. Although there is a clear need for replication in measured gene studies, it is our hope that by using haplotypes investigators will be able to clarify inconsistencies in the gene × environment literature.

Contact Michael J. Sulik