Heritability and polygenic load for comorbid anxiety and depression

Abstract

Anxiety and depression commonly occur together resulting in worse health outcomes than when they occur in isolation. We aimed to determine whether the genetic liability for comorbid anxiety and depression was greater than when anxiety or depression occurred alone. Data from 12,792 genotyped twins (ages 38-85) were analysed, including 1,986 complete monozygotic and 1,594 complete dizygotic pairs. Outcomes were prescription of antidepressant and anxiolytic drugs, as defined by the World Health Organization Anatomical Therapeutic Chemical Classification System (ATC) convention, for comorbid anxiety and depression (n = 1028), anxiety only (n = 718), and depression only (n = 484). Heritability of each outcome was estimated using twin modelling, and the influence of common genetic variation was assessed from polygenic scores (PGS) for depressive symptoms, anxiety, and 40 other traits. Heritability of comorbid anxiety and depression was 79% compared with 41% for anxiety and 50% for depression alone. The PGS for depressive symptoms likewise predicted more variation in comorbid anxiety and depression (adjusted odds ratio per SD PGS = 1.53, 95% CI = 1.43-1.63; ΔR² = 0.031, ΔAUC = 0.044) than the other outcomes, with nearly identical results when comorbid anxiety and depression was defined by International Classification of Diseases (ICD) diagnoses (adjusted odds ratio per SD PGS = 1.70, 95% CI = 1.53-1.90; ΔR² = 0.036, ΔAUC = 0.051). Individuals in the highest decile of PGS for depressive symptoms had over 5 times higher odds of being prescribed medication for comorbid anxiety and depression compared to those in the lowest decile. While results on a predominant role of depressive symptoms may have been biased by the size and heterogeneity of available data bases, they are consistent with the conclusion that genetic factors explain substantially more variation in comorbid anxiety and depression than anxiety or depression alone.

Fig. 1. Proportion of outcome groups with prescription divided by drug class.

Note: N06AA-AB = non-selective monoamine reuptake inhibitors, selective serotonin reuptake inhibitors (SSRIs); N06AC-AX = monoamine oxidase inhibitors (non-selective), monoamine oxidase A inhibitors, other antidepressants (which include certain selective serotonin and noradrenaline reuptake inhibitors as well as other drugs); Benzo = benzodiazepine derivatives; BusProPre = Buspirone, Propranolol, Pregabalin. Drug classifications are based on their Anatomical Therapeutic Chemical code (ATC).

Fig. 2. Genetic liability for anxiety and depression outcomes.

A Twin heritability for each outcome (AE-model), error bars represent 95% CI. B Accuracy (ΔR²) of depressive symptoms PGS (DEP-PGS) from the Polygenic Index Repository in predicting each outcome above the baseline model.

Fig. 3. Results from regression models with single PGS predictors including covariates.

Adjusted OR per SD increase in PGSs predicting prescription-based outcomes. Error bars represent 95% CI.

Fig. 4. Genetic risk across the PGS distribution.

Adjusted OR per SD PGS with 95% CI for each decile of Depressive symptoms PGS (DEP-PGS) compared with the first decile.