下丘脑-垂体-肾上腺轴是攻击产生与发展的关键生理基础，其功能与攻击呈非线性关系模式，该非线性模式可能受环境调控并通过行为控制系统间接发挥作用。对530名成年早期被试（19.35 ± 1.59岁, 53.4%女生）进行间隔一年的追踪研究，采用多基因累加风险得分范式考察HPA轴基因与攻击间的非线性关系，并探索亲子关系的调节作用与冲动性的中介作用。结果发现：（1）亲子冲突而非亲子亲合能够与HPA轴基因二次项交互影响攻击，只有经历高水平的亲子冲突时，HPA轴基因与攻击呈U型关系；（2）同时性和纵向分析一致发现，冲动性在基因—亲子冲突交互作用与攻击间发挥中介作用，仅在纵向模型中基因—亲子冲突交互作用对攻击的直接效应显著。研究结果为发展精神病理学的“多因一果”现象提供了支持，并且为揭示神经生理应激系统影响攻击的复杂作用机制提供了新的研究视角。

Aggression, which is a common and remarkably damaging problem, has its roots in stress-responsive systems. Although children display a wide range of individual differences in the hypothalamic-pituitary-adrenal (HPA) axis function, numerous initial conditions of stress reactivity may reach the same end state. In other words, both hyper- and hypo-responses to stress tend to induce heightened risks for aggressive behavior. Informed by the concept of equifinality in developmental psychopathology, this study, through its focus on hypothalamic-pituitary-adrenocortical (HPA) genetic susceptibility, examines the nonlinear (U-shaped) relationship between stress responsiveness and the severity of aggression. Additionally, since it remains unclear whether the equifinality phenomenon is evident in distal behavioral phenotypes or proximal endophenotypes, this study also evaluated, on the basis of “psychobiological model of antisocial behavior” framework, a mediated moderation model to examine the effects of the HPA axis genes on aggression, as moderated by the parent-child relationship and mediated by impulsivity.

A total of 530 participants (mean age 19.35 ± 1.59 years at Time 1, 53.4% females) completed two assessments at a one-year interval. Data on self-reported aggression, parent-child relationship, and impulsivity were collected, and DNA was extracted from saliva. All of the measures exhibited good reliability. The multilocus genetic profile score (MGPS) was calculated using four polymorphisms within HPA axis-related genes, namely NR3C2 gene rs2070951 polymorphism, CRHR1 gene rs110402 polymorphism, COMT gene rs4680 polymorphism, and BDNF gene rs6265 polymorphism. For each participant, genotyping of the four HPA axis genes was performed using improved multiplex ligation detection reaction. This was followed by a series of hierarchical regressions that were conducted to examine the U-shaped relationship between MGPS and aggression, and the moderating role of parent-child relationships and the mediating role of impulsivity was tested in a mediated moderation model. To test the robustness of the results, a series of sensitivity analysis were conducted. Specifically, the mediated moderation models of each polymorphism were examined to explain the power of MGPS approach. Besides, an internal replication and meta-analysis were conducted by randomly splitting the total sample into two subsamples.

The U-shaped relation between HPA MGPS and aggression was not observed. However, a quadratic, U-shaped relationship was observed between the additive genetic risk of HPA and aggression. Both low and high MPGS carriers exhibited high levels of aggression when exposed to higher levels of parent-child conflict. Nevertheless, parent-child cohesion did not exhibit such moderating effects. Furthermore, the moderating effect was mediated by impulsivity. Similarly, there was a significant quadratic relationship between MGPS and impulsivity when parent-child conflicts were at comparatively high levels, with results displaying a U-shaped relationship. However, an inverted U-shaped relationship between MGPS and impulsivity was observed when parent-child conflicts were relatively lower. Impulsivity was a significant risk factor for aggression, with high impulsivity predicting high levels of aggression. The sensitive analysis revealed that the additive genetic risk of HPA accounted for a more significant effect than any single gene. The mediated moderation model was replicated in both two subsamples.

These findings inform our understanding of how additive genetic variants in the HPA axis and its response to adversity are involved in the etiology of aggressive behavior. It is likely that, because of the U-shaped relationship, the association between HPA genetic function and aggression is more complex than what the general perspectives — “the more the genetic risk variants, the higher the likelihood of aggressive behavior.” Moreover, these findings provide support for the phenomenon of equifinality in developmental psychopathology. Domain-specific findings regarding the differences in parent-child conflicts and the cohesion model imply that the U-shaped function of HPA genes cannot be generalized to positive environmental influences.