Abstract
Depressive and anxiety disorders are among the leading causes of disability worldwide. Therapist-guided, Internet-delivered cognitive behaviour therapy (ICBT) is an established treatment for depression and anxiety, but a considerable proportion of treated patients do not achieve sufficient improvement. Predicting symptom change from clinical variables alone is challenging. Genetic data could potentially add predictive power and help us understand who will benefit most from ICBT. We conducted a study including 2668 adults (62% women, mean age 35.6 years) from the Swedish MULTI-PSYCH cohort to investigate the association between polygenic risk scores (PRS) from eight psychiatric and cognitive phenotypes and symptom change after ICBT. All participants had been diagnosed with depression, panic disorder or social anxiety disorder and treated with ICBT. The primary clinical outcome was a harmonised score across three different diagnosis-specific symptom rating instruments and measured weekly throughout ICBT. PRS were computed for attention-deficit/hyperactivity disorder, autism spectrum disorder, bipolar disorder, major depressive disorder and schizophrenia, cross-disorder psychopathology, educational attainment, and intelligence, using large discovery data sets. Linear mixed-effects models identified a significant association between the PRS for educational attainment (PRS-EDU) and symptom change (B = -0.73, p = .03), suggesting that a higher PRS-EDU was associated with lower symptom severity. This remained significant after additional covariate adjustment. No other PRS were significant. In the adjusted model, there was a significant PRS-EDU*time interaction, indicating that PRS-EDU also influenced the symptom change rate during treatment. When excluding outliers, the interaction effect was significant in both the crude and adjusted model. While these results await replication, they could have important implications for how the ICBT could be adapted to suit a wider portion of the population.