High genetic loading of schizophrenia predicts poor response to lithium in patients with bipolar disorder: A polygenic score and cross-trait genetic analysis

ABSTRACT
Importance Lithium is a first-line mood stabilizer for the maintenance treatment of Bipolar Disorder (BPD). However, the efficacy of lithium varies widely, with a non-response rate of up to 30%. Biological response markers and predictors are lacking.
Objective Genetic factors are thought to mediate lithium treatment response, and the previously reported genetic overlap between BPD and schizophrenia (SCZ) led us to test whether a polygenic score (PGS) for SCZ could predict lithium treatment response in BPD. Further, we explored the potential molecular underpinnings of this association.
Design Weighted SCZ PGSs were computed at ten p-value thresholds (PT) using summary statistics from a genome-wide association study (GWAS) of 36,989 SCZ cases, and genotype data for BPD patients from the Consortium on Lithium Genetics (ConLi+Gen). For functional exploration, we performed a cross-trait meta-GWAS and pathway analysis, combining GWAS summary statistics on SCZ and lithium treatment response.
Setting International multicenter GWAS.
Participants Patients with BPD who had undergone lithium treatment were genotyped and retrospectively assessed for long-term treatment response (n=2,586).
Main outcome measures Clinical treatment response to lithium was defined on both the categorical and continuous scales using the ALDA score. The effect measures include odds ratios (ORs) and the proportion of variance explained (R2), and a significant association was determined at p<0.05.
Results The PGS for SCZ was inversely associated with lithium treatment response in the categorical outcome (p=8×10−5), at PT <5×10−2. Patients with BPD who had low polygenic load for SCZ responded better to lithium, with ORs for lithium response ranging from 3.46 [95%CI: 1.42-8.41 at 1st decile] to 2.03 [95%CI: 0.86-4.81 at the 9th decile], compared to the patients in the 10th decile of SCZ risk. In the cross-trait meta-GWAS, 15 genetic loci that may have overlapping effects on lithium treatment response and susceptibility to SCZ were identified. Functional pathway and network analysis of these loci point to the HLA complex and inflammatory cytokines (TNFα, IL-4, IFNγ) as molecular contributors to lithium treatment response in BPD.
Conclusions and Relevance The study provides, for the first-time, evidence for a negative association between high genetic loading for SCZ and poor response to lithium in patients with BPD. These results suggest the potential for translational research aimed at personalized prescribing of lithium.
Question Does a polygenic score for Schizophrenia (SCZ) predict response to lithium in patients with Bipolar Disorder (BPD)? What are the molecular drivers of the association between SCZ and lithium treatment response?
Findings We found an inverse association between genetic loading for SCZ risk variants and response to lithium in patients with BPD. Genetic variants in the HLA region on chromosome 6, the antigen presentation pathway and markers of inflammation (TNFα, IL-4, IFNγ) point to molecular underpinnings of lithium treatment response in BPD.
Meaning In patients with BPD, an assessment of a polygenic load for SCZ risk variants may assist in conjunction with clinical data to predict whether they would respond to lithium treatment.
Subject Area
- Biochemistry (9152)
- Bioengineering (6789)
- Bioinformatics (24037)
- Biophysics (12142)
- Cancer Biology (9550)
- Cell Biology (13808)
- Clinical Trials (138)
- Developmental Biology (7649)
- Ecology (11719)
- Epidemiology (2066)
- Evolutionary Biology (15522)
- Genetics (10654)
- Genomics (14337)
- Immunology (9495)
- Microbiology (22872)
- Molecular Biology (9113)
- Neuroscience (49070)
- Paleontology (355)
- Pathology (1485)
- Pharmacology and Toxicology (2572)
- Physiology (3851)
- Plant Biology (8341)
- Synthetic Biology (2299)
- Systems Biology (6199)
- Zoology (1302)