Multi-omics analysis links kynurenic acid to lithium response in bipolar disorder

The kynurenine pathway (KP) is implicated in bipolar disorder through its effects on glutamate signaling and neuroinflammation, but its role in clinical heterogeneity and treatment response is unclear. We used a multi-omic approach to test whether KP metabolites and their corresponding genetic risk scores are associated with clinical phenotypes, particularly response to lithium. The study included 457 patients with bipolar disorder (63.2% female, mean age=38.94 ± 14.85 years) from Mayo Clinic Bipolar Disorder Biobank. Serum levels of KP metabolites (tryptophan, kynurenine, kynurenic acid [KYNA], quinolinic acid [QUIN], and picolinic acid [PIC]) were measured using ultra-high-performance liquid chromatography. Polygenic scores (PGS) for these metabolites were constructed using public GWAS summary statistics. Multivariable regression models were used to examine the relationships between metabolite levels, PGS and clinical features, including manic psychosis, early age of onset, suicide attempt, and lithium response (Alda A scale). Higher tryptophan and PIC were nominally associated with decreased odds of a lifetime suicide attempt history (OR = 0.73, p = 0.025; and OR = 0.76, p = 0.030, respectively), while higher KYNA was associated with increased odds (OR = 1.51, p = 0.008). Higher KYNA was nominally positively associated with lithium response (β = 0.56, p = 0.021). However, these preliminary metabolite-phenotype associations did not survive correction for multiple comparisons. No significant associations were observed for manic psychosis or early AOO. In the genetic analysis, only PGS for KYNA and PIC predicted corresponding serum metabolite levels. Notably, PGS-KYNA was significantly associated with better lithium response (β = 0.57, p = 0.007) and was the only finding to remain significant after multiple-testing corrections (adjusted p = 0.042). Furthermore, in a joint model, both higher serum KYNA levels (β = 0.59, p = 0.008) and PGS-KYNA (β = 0.53, p = 0.011) independently predicted better lithium response. These findings reinforce the KP as a potential therapeutic target in bipolar disorder and support multi-omic approaches to advance precision psychiatry.