A genome-wide association study (GWAS) identified the first genomic locus and developed a polygenic risk score (PRS) for predicting lithium total body clearance (CLLi). These findings were published in The Lancet Psychiatry.
This study analyzed data from 2 clinical cohorts of patients living in Sweden taking lithium for bipolar disorder (BD). The primary outcome was the natural logarithm of daily CLLi, defined as the ratio between daily intake and serum concentration. A GWAS for CLLi was performed to identify single nucleotide polymorphisms (SNPs) independently associated with CLLi and a PRS was formulated using the SNP risk effects.
Cohorts 1 (n=584) and 2 (n=1773) were 62% and 60% women, aged mean 51.8 (range, 18-89) and 54.2 (range, 17-85) years (P =.0005), 81.0% and 96.4% used lithium sulphate (P <.0001), mean daily lithium intake was 24.3 (range, 6.0-52.4) and 21.6 (range, 6.0-52.8) mmol (P <.0001), mean serum concentration was 0.7 (range, 0.4-0.9) and 0.6 (range, 0.4-0.9) mmol/L (P <.0001), and CLLi was 3.5 (range, 2.01-4.56) and 3.5 (range, 2.07-4.42) log(L/day) (P =.002), respectively.
The GWAS identified 2 SNPs in 1 locus. The lead SNP, rs583503 (P =3.8×10-8), was located on chromosome 11 in the intron of the long noncoding RNA Lnc-FAM118B-1.
After accounting for age, gender, and the first 4 principal components of CLLi as fixed effects, carrying 1 additional copy of the minor allele (G) at rs583503 associated with a decrease in CLLi (β, -0.53).
Using the risk effects of the most significant SNPs, a small, but significant proportion of the variance in CLLi was explained by the generated PRS (P =.0046).
PRSs for BMI and estimated glomerular filtration rate (eGFR) were positively associated and the PRS for blood urea nitrogen was negatively associated with the individual-level effects of CLLi accounting for age and gender. No association was observed between the PRSs for CLLi and BD.
To determine whether including genetic data could improve the prediction of CLLi, a baseline model which included clinical predictors (age, gender, log [eGFR], use of diuretics, renin–angiotensin–aldosterone system-acting drugs, serum lithium) was compared with the model including genetics. The predictive power was only marginally increased in both cohorts 1 (59.32% vs 59.36%) and 2 (49.21% vs 50.03%) with the addition of genetic data, respectively.
The major limitation of this study was the lack of robust data on actual lithium intake. In order to evaluate accurate intake, a point-of-care with regular monitoring approach may be needed.
“This is the largest study […] on lithium population pharmacokinetics and the first to combine its results with a GWAS. We provide data that support the findings from previous small cohorts, extend the literature on how several known predictors affect lithium pharmacokinetics, and thus provide a model that explains a significant amount of variance in CLLi. Furthermore, we provide evidence for 1 locus on chromosome 11 to be associated with CLLi at genome-wide significance, as well as associations with several PRSs, suggesting a novel avenue for individualized medicine in lithium treatment strategies,” concluded the study authors.
Disclosure: Multiple authors declared affiliations with industry. Please refer to the original article for a full list of disclosures.
Millischer V, Matheson GJ, Bergen SE, et al. Improving lithium dose prediction using population pharmacokinetics and pharmacogenomics: a cohort genome-wide association study in Sweden. Lancet Psychiatry. 2022;9(6):447-457. doi:10.1016/S2215-0366(22)00100-6