Transcriptomic signatures of lithium treatment in peripheral blood mononuclear cells from bipolar disorder patients

Background

Bipolar disorder is a chronic and severe psychiatric condition affecting approximately 2% of the global population, characterised by recurrent episodes of depression and mania and increased suicide risk if untreated [1]. The most effective mood stabilizer remains lithium, preventing both manic and depressive states and reducing suicide risk, even if not all patients are fully responsive [2]. Despite its widespread use, the transcriptomic effects of lithium have not yet been explored. Therefore, peripheral blood mononuclear cells (PBMCs) can provide a valuable and accessible model to investigate the systemic molecular changes associated with psychiatric treatments.

Aims

We aimed to identify peripheral transcriptomic signatures associated with lithium treatment in patients with bipolar disorder and to characterise multi-level molecular differences across genes, pathways, and gene co-expression networks.

Methods

A cohort of 116 patients with bipolar disorder, recruited at IRCCS Ospedale San Raffaele, was analysed using bulk RNA sequencing of PBMCs. Participants were classified as lithium-treated (n = 66) if they had been treated for at least six months. The comparison group (n = 50) included individuals who never took lithium, those treated for less than six months, and those who had discontinued it. Differential gene expression was performed using edgeR, adjusting for sex, and association with age was assessed using a linear model. Pathway-level activity was estimated through gene set enrichment analysis (GSEA) and gene set variation analysis (GSVA), using Gene Ontology Biological Process and Hallmark gene sets. Cellular composition was estimated via CIBERSORTx deconvolution. Additionally, weighted gene co-expression network analysis (WGCNA) was performed to identify gene co-expression modules, which were tested for association with lithium treatment.

Results

Differential gene expression analysis identified three upregulated genes in lithium-treated individuals (FDR < 0.05): SEPTIN11, DGKK, and TPPP3. After excluding from the analysis seven individuals who were taking lithium for less than six months, two genes were found to be downregulated (FDR < 0.05): LEF1-AS1 and PLXDC1. Pathway-level analysis revealed a significant modulation of multiple biological pathways, with positive enrichment of IL2–STAT5 and TNFα signaling, and negative enrichment of interferon gamma response, interferon alpha response, and MYC targets (V2). GSVA-based comparison further highlighted an increased activity of pathways related to lithium treatment, including brain morphology and cardiovascular conduction. CIBERSORTx deconvolution showed decreased proportion of naïve T cells in lithium-treated individuals (FDR = 0.027), supporting an effect on immune cell composition. Finally, WGCNA identified a platelet activation-related module significantly associated with lithium exposure (r = 0.25, FDR = 0.01), suggesting involvement of vascular and inflammatory processes.

Conclusions

These findings suggest that lithium treatment is associated with coordinated transcriptomic and immunological changes in peripheral blood, involving both inflammatory pathway modulation and cellular composition. Multi-level molecular profiling, from genes to pathways and gene coexpression networks, captures a biological variability associated with lithium exposure and may provide a peripheral approximation of its systemic effects. Exploratory analyses further suggest a potential link between peripheral transcriptomic features and measures of brain integrity.

References

Goes FS. Diagnosis and management of bipolar disorders. BMJ. 12 aprile 2023;381:e073591. doi:10.1136/bmj-2022-073591
Gitlin M, Bauer M. Lithium: current state of the art and future directions. Int J Bipolar Disord. 28 novembre 2024;12(1):40. doi:10.1186/s40345-024-00362-7