Polygenic risk scores of lithium response and treatment resistance in major depressive disorder

Abstract

Treatment response and resistance in major depressive disorder (MDD) are suggested to be heritable. Due to significant challenges in defining treatment-related phenotypes, our understanding of their genetic bases is limited. This study aimed to derive a stringent definition of treatment resistance and to investigate the genetic overlap between treatment response and resistance in MDD. Using electronic medical records on the use of antidepressants and electroconvulsive therapy (ECT) from Swedish registers, we derived the phenotype of treatment-resistant depression (TRD) and non-TRD within ~4500 individuals with MDD in three Swedish cohorts. Considering antidepressants and lithium are first-line treatment and augmentation used for MDD, respectively, we generated polygenic risk scores (PRS) of antidepressants and lithium response for individuals with MDD and evaluated their associations with treatment resistance by comparing TRD with non-TRD. Among 1778 ECT-treated MDD cases, nearly all (94%) used antidepressants before their first ECT and the vast majority had at least one (84%) or two (61%) antidepressants of adequate duration, suggesting these MDD cases receiving ECT were resistant to antidepressants. We did not observe a significant difference in the mean PRS of antidepressant response between TRD and non-TRD; however, we found that TRD cases had a significantly higher PRS of lithium response compared to non-TRD cases (OR = 1.10-1.12 under various definitions). The results support the evidence of heritable components in treatment-related phenotypes and highlight the overall genetic profile of lithium-sensitivity in TRD. This finding further provides a genetic explanation for lithium efficacy in treating TRD.

Fig. 1. The results of lithium response PRS and TRD.

A The association between PRS of lithium response and TRD. ORs for TRD associated with per-SD increase in the PRS of lithium response. Each panel shows ORs under different TRD definitions (see the “Materials and methods” section). B The proportion of variance in MDD treatment resistance explained by PRS of lithium response (Nagelkerke’s R² on the liability scale). C The proportion of MDD cases (TRD and non-TRD) in each quartile (Q1–Q4) of the PRS of lithium response. Each panel represents one definition of TRD and non-TRD (see the “Materials and methods” section). The P value for trend analysis was 0.0006, 0.0015, and 0.0036 under broad, narrow_1, and narrow_2 definitions, respectively. “None”: model adjusted only for the first four PCs. | “MDD_adj”: model additionally adjusted for the PRS of MDD. | “BIP_adj”: model additionally adjusted for the PRS of bipolar disorder. | “BIP_MDD_adj”: model additionally adjusted for the PRS of both MDD and BIP.