Lithium for Bipolar Disorder and Risk of Thyroid Dysfunction and Chronic Kidney Disease

Abstract

Importance: Literature indicates adverse effects of lithium on thyroid and kidney function. However, existing data are heterogeneous, with limitations in quality and lack evaluation of adverse effects of lithium vs other mood stabilizers, especially commonly prescribed second-generation antipsychotics. Lithium serum level thresholds associated with thyroid and kidney abnormalities remain unknown.

Objective: To examine risk of thyroid and kidney dysfunction in patients with incident bipolar disorder (BD) treated with lithium and other mood stabilizers and antipsychotics in an Asian population and to determine lithium serum level cutoffs associated with these physical complications.

Design, setting, and participants: This population-based retrospective cohort study identified patients aged 15 years or older with first-diagnosed BD in Hong Kong from 2002 to 2018, utilizing a medical record database of public health care services. Data analysis was performed from February to May 2024.

Exposures: Lithium vs nonlithium treatment. The nonlithium group was further stratified into valproate, olanzapine, quetiapine, and risperidone groups.

Main outcomes and measures: Main outcomes were hypothyroidism, hyperthyroidism, and chronic kidney disease stage 3 or higher (CKD3+), with additional investigation on CKD stage 4 or higher (CKD4+) and end-stage kidney disease (ESKD). Outcomes were ascertained using laboratory test results. Cox proportional hazards regression analyses were performed for risk estimation with adjusted hazard ratios (aHRs) and 95% CIs. Receiver operating characteristic analyses with the Youden index were employed to determine lithium serum level cutoffs associated with thyroid and kidney dysfunction.

Results: There were 4752 individuals with analyzable data for hypothyroidism (mean [SD] age, 39.5 [15.6] years; mean [SD] follow-up, 8.4 [4.8] years; 2889 female [60.8%]), 4500 with data for hyperthyroidism (mean [SD] age, 39.7 [15.6] years; mean [SD] follow-up, 8.7 [4.7] years; 2716 female [60.4%]), and 7029 with data for CKD (mean [SD] age, 37.9 [14.8] years; mean [SD] follow-up, 8.3 [4.8] years; 4251 female [60.5%]). Lithium was associated with increased risk of hypothyroidism (aHR, 2.00; 95% CI, 1.72-2.33) and CKD3+ (aHR, 1.35; 95% CI, 1.15-1.60), but not CKD4+ or ESKD, compared with nonlithium treatments. Higher lithium serum levels were associated with elevated rates of hypothyroidism (aHR, 2.08; 95% CI, 1.67-2.59), hyperthyroidism (aHR, 1.81; 95% CI, 1.31-2.50), and CKD3+ (aHR, 2.11; 95% CI, 1.57-2.85). Greater number of lithium toxicity episodes was associated with increased CKD3+ risk. Valproate, olanzapine, quetiapine, and risperidone generally exhibited reduced likelihood of thyroid dysfunction and CKD3+ compared with lithium, without any difference in advanced CKD. Mean lithium serum levels greater than 0.5028 mEq/L, greater than 0.5034 mEq/L, and greater than 0.5865 mEq/L represented thresholds associated with hypothyroidism, hyperthyroidism, and CKD3+, respectively.

Conclusions and relevance: In this cohort study of patients with incident BD, lithium was associated with a mildly increased risk of thyroid dysfunction and CKD in a predominantly Chinese population. The identified lithium level thresholds associated with risks of physical complications may facilitate the development of evidence-based guidelines recommending lithium treatment, particularly in Asian populations, and the promotion of personalized care and risk-benefit balancing in the treatment for BD.

Figure 1.. Lithium Use and Risk of Hypothyroidism, Hyperthyroidism, and Chronic Kidney Disease (CKD) in Patients With Bipolar Disorder

aHR indicates adjusted hazard ratio. ^aBonferroni correction for multiple comparison was applied (corrected P value = .002). ^bResults remained significant after Bonferroni correction.

Figure 2.. Lithium Treatment Characteristics and Risk of Hypothyroidism, Hyperthyroidism, and Chronic Kidney Disease (CKD) in Patients With Bipolar Disorder

The median value of mean lithium serum level of lithium users included for hypothyroidism, hyperthyroidism, and CKD outcomes were 0.51 mEq/L, 0.50 mEq/L, and 0.52 mEq/L, respectively (to convert to millimoles per liter, multiply by 1). The median value of cumulative lithium exposure duration of lithium users included for hypothyroidism, hyperthyroidism, and CKD outcomes were 1.95 years, 2.45 years, and 2.47 years, respectively. The median value of number of lithium toxicity episodes for the 3 study outcomes (>1.0 mEq/L, >1.2 mEq/L, and >1.5 mEq/L) were all 0. aHR indicates adjusted hazard ratio. ^aBonferroni correction was applied for multiple comparison (threshold for P value = .003). ^bRegression models adjusted for age at bipolar disorder diagnosis, sex, hypertension, dyslipidemia, diabetes, age-adjusted Charlson Comorbidity Index score, and prescriptions of other antipsychotics (ie, antipsychotics other than olanzapine, quetiapine, and risperidone), other mood-stabilizing anticonvulsants (ie, carbamazepine or lamotrigine), any antidepressants, and studied mood stabilizers other than the specified agent. ^cResults remained significant after Bonferroni correction. ^dRegression models adjusted for age at bipolar disorder diagnosis, sex, hypertension, hyperlipidemia, diabetes, age-adjusted Charlson Comorbidity Index score, catchment area of psychiatric service receipt, and prescriptions of other antipsychotics, mood-stabilizing anticonvulsants, antidepressant, nonsteroid anti-inflammatory drugs, angiotensin converting enzyme inhibitors or angiotensin II receptor blocks, and diuretics.

Figure 3.. Receiver Operating Characteristic Analysis on Mean Lithium Serum Levels Associated With Occurrence of Adverse Outcomes

Mean lithium serum levels greater than 0.5028 mEq/L (sensitivity = 0.67; specificity = 0.52), greater than 0.5034 mEq/L (sensitivity = 0.62; specificity = 0.52), and greater than 0.5865 mEq/L (sensitivity = 0.53; specificity = 0.67) was associated with occurrence of hypothyroidism (A), hyperthyroidism (B), and chronic kidney disease stage 3 (C), respectively. To convert to millimoles per liter, multiply by 1.