Unveiling genetic architectures for stratifying trajectories of adolescent depression

Adolescent-onset depression is characterised by high levels of inter-individual variability and genetic heritability. Investigating the genetic factors that underlie trajectories of depression is crucial to enhancing mechanistic understanding of the onset, persistence and severity of adolescent depression. However, predicting trajectories from complex and heterogeneous genetic architectures in psychopathology poses challenges due to the high genetic correlation among traits. It remains unclear whether multi-trait models provide a superior prediction of depression trajectories compared to univariate models focused solely on depression. Addressing this question is important for effective stratification and targeted treatments. To validate depression trajectories during adolescence, we conducted growth mixture modelling in two longitudinal cohorts (ABCD and ALSPAC; total N=20,509). We then computed polygenic risk scores for seven traits: major depressive disorder (MDD), anxiety, neuroticism, schizophrenia, bipolar disorder, attention-deficit-hyperactivity disorder (ADHD), and autism for participants with European ancestry. We also generated MDD scores for individuals of African, East Asian, and Hispanic ancestries in ABCD. Using genomic structural equation modelling, we compared three multi-trait factor models (common, correlated, hierarchical) to assess the relationships among the seven traits. To generate multi-trait risk scores from these models in our target cohorts, we conducted multivariate genome-wide association analysis to determine the effects of single nucleotide polymorphisms on genetic latent p-factors. Finally, we examined the association between all polygenic risk scores for univariate traits and multi-trait models with depression trajectories. Four distinct trajectories were replicated across two cohorts with partial age-range overlap encompassing adolescents from two generations. Trajectories included stable low, adolescent acute, increasing and decreasing. The hierarchical factor model was the best fit for multi-trait genetic information and was most predictive of adolescent acute trajectories (odds ratio [OR], 1.46; 95% CI, 1.27-1.68), with increasing and decreasing showing comparable risk (OR, 1.32; 95% CI, 1.16-1.50). Multi-trait models showed a similar genetic risk for depression trajectory as MDD-only risk across trajectories. Anxiety was associated with the adolescent acute trajectories (OR, 1.27; 95% CI, 1.10-1.45). Psychotic conditions were associated with later-onset depression patterns and ADHD with earlier-onset, aligning with the developmental timelines of these respective conditions. The investigated genetic traits collectively contribute to diverse longitudinal patterns of depression, varying in severity, onset, and duration. A hierarchical factor model of multivariate genetic psychopathology demonstrated a comparable prediction of genetic risk to univariate depression scores for stratifying longitudinal depression in adolescence. It is important to acknowledge the genetic influence of multiple conditions on depression outcomes, particularly at different stages of vulnerability. Taking into account detailed and integrated genetic information will be valuable in effectively stratifying trajectories across adolescence and mental health conditions.