Abstract
Introduction
Bipolar disorder (BD) is a chronic medical condition characterized by episodes of mania, hypomania, and intertwined depression. 1 The lifetime prevalence of BD ranges between 1% and 2% in the global population. 2 In bipolar I disorder, the rate of lifetime psychiatric comorbidities such as alcohol and substance use disorders, anxiety disorders, post-traumatic stress disorder (PTSD), personality disorders, and eating disorders has been reported to range from 50% to 70%. 3 Longitudinal studies have shown a suicide completion rate of 3–10% among BD patients. 4 In addition, many medical comorbidities, whether coexisting with BD or arising from treatment, occur at rates higher than in the general population. 3 Unfortunately, treatment resistance is rather common, with up to 33% of patients suffering from refractory BD. 5 The associated burden in the US (comprising cost of illness, cost of direct and indirect healthcare utilization, somatic comorbidity, mortality, and loss of workplace productivity) exerted on patients and society was estimated at $194.8 billion per year in 2009. 6 Furthermore, BD patients exhibit a higher rate of hospitalizations. 7
Lithium prevails as the gold standard treatment for BD, particularly for mania and maintenance therapy. 8 Notably, data from Japan shows that between January 2005 and March 2014, 136,956 lithium prescriptions were issued to 5823 patients, among whom 19.5% were BD patients. 9 However, lithium remains underprescribed for BD compared to atypical antipsychotics. Indeed, a study reporting on the data from the Global Bipolar Cohort collaborative network showed that lithium was prescribed in 29% of patients, while anticonvulsants were prescribed in 44%, atypical antipsychotics in 42%, and antidepressants in 38%. 10 Lithium is FDA-approved for the treatment of bipolar I disorder, specifically for the management of acute mania and maintenance therapy. Lithium has a unique antisuicidal property and proven efficacy in protecting against both depressive and manic episodes in patients with bipolar disorder. 11 In addition, lithium has proven effective in the treatment of acute depression and treatment-resistant depression,12,13 with superior efficacy in preventing relapse of major depression and suicidality compared to second-generation antipsychotics. 14
However, since the early 1970s, reports have emerged on the adverse effects of lithium use on renal function. 15 Concerns regarding nephrotoxicity led to a decline in the use of lithium treatment for BDI despite its effectiveness. 15 Three categories of lithium nephrotoxicity can be distinguished: acute lithium intoxication, nephrogenic diabetes insipidus (NDI), and chronic kidney disease (CKD).16,17 Indeed, renal biopsy studies document that chronic lithium treatment is associated with morphological changes such as chronic focal atrophy with interstitial fibrosis, tubular degeneration, and glomerular sclerosis.18–21 Moreover, lithium is postulated to enter principal cells of renal collecting ducts via amiloride-sensitive epithelial sodium channel (ENaC). 22 This results in the downregulation of aquaporin 2 (AQP2) at the mRNA level 23 through decreased intracellular calcium signaling that thereby inhibits glycogen synthase kinase 3 (GSK3). 24 This lithium-driven reduction in AQP2 function has been shown in murine models 24 and aligns with the clinically observed polyuria related to NDI, as these proteins allow for the passage of water along its osmotic gradient, thus increasing urine concentration. 24 A meta-analysis including comparative and non-comparative studies demonstrated a significant prevalence of impaired kidney function in patients undergoing long-term lithium therapy, suggesting that a quarter of cases could develop advanced CKD. 25 In contrast, a recent Swedish study using the new-user active-comparator design and comprising a large cohort of 10,946 patients, 5308 of whom were undergoing lithium therapy and 5638 on valproate therapy, has shown no elevated 10-year CKD risk with lithium therapy compared to valproate. 26 In a cross-sectional cohort study including 2334 patients, Fransson et al. 27 reported that patients with BD or schizoaffective disorder being treated with lithium had a steeper estimated Glomerular filtration rate (eGFR) decline compared to the reference population, although the trajectory of kidney function decline varied widely, with the effect of lithium showing great inter-individual variability. However, patients taking lithium for extended durations undergo normal aging, which is associated with renal senescence. 28 In addition, studies have shown that affective disorder diagnosis, regardless of treatment could be associated with kidney function decline.29,30 Furthermore, BD seems to be linked to higher rates of various physical comorbidities, including diabetes and cardiovascular disease. 31 Such confounders must therefore be matched for at baseline, as they would otherwise bias studies in favor of inflated lithium nephrotoxicity levels. As such, the debate about lithium-induced versus lithium-associated nephropathy due to an underlying affective disorder diagnosis remains under intense investigation. To address this issue, we performed a meta-analysis of studies that assessed renal function in patients with affective disorders, especially BDs, on lithium therapy compared to studies of patients not on lithium therapy.
Methods
Eligibility criteria for considering studies for this review
Eligible studies must have met all the following inclusion criteria: observational studies—prospective, retrospective and cross-sectional studies of adult patients with diagnoses of affective disorders including BD on lithium, with control groups not on lithium (including patients on other mood stabilizers, second-generation antipsychotics, or on no medications for affective disorders); reported at least one of the following outcomes: eGFR, serum creatinine, creatinine clearance, new albuminuria, and new or worsening CKD. Studies without BD patients and studies that only included CKD patients were excluded. Case reports, case series, abstracts, conference abstracts, reviews, and articles not reported in English were excluded from the study.
Search strategy for identifying studies
The review followed the PRISMA guidelines 32 (Supplemental Table 1s: PRISMA 2020 checklist).
A comprehensive search of several databases from inception to October 2023 was conducted for English language publications. The databases included Cochrane, Embase, PubMed, Scopus, Web of Science, and CINAHL. The search strategy was designed and conducted by an experienced librarian with input from the study’s principal investigator. Controlled vocabulary supplemented with keywords was used to search for studies describing lithium nephrotoxicity in patients with affective disorders. Figure 1(a) shows the PRISMA flowchart and outlines the search strategy, listing all the search terms used. This review was registered prospectively with PROSPERO (CRD42024517414).
Study selection process and quality assessment of included studies. (a) PRISMA flow chart depicting the study selection process, including the number of records identified, screened, excluded, and included in the final analysis. (b) Risk of bias and quality assessment of included studies, summarizing methodological strengths and limitations.
Study selection
Three independent assessors (MMM, MAR, RM) conducted article screening and data extraction using the Covidence software (Covidence.org, Software Development, Melbourne, VIC, Australia). Any disagreements were adjudicated (MMM) and discussed with co-authors as necessary.
Reported outcomes
Estimated glomerular filtration rate outcomes were reported in mean (ml/min/1.73 m2) and annual decline in eGFR (ml/min/1.73 m2). Mean serum creatinine (mg/dl) and mean creatinine clearance (ml/min) were retrieved. The number of new CKD or CKD progression and new albuminuria events was also retrieved.
Data collection and risk of bias assessment
The following participant characteristics were extracted: mean age (years) and the number and percentage of females. The following baseline characteristics were extracted: lithium exposure duration (years), lithium blood levels (mEq/l), baseline eGFR (ml/min/1.73 m2), and CKD (
Data synthesis and analysis
Pooled mean differences and proportions of our data were analyzed using the random-effects, generic inverse variance method for continuous data and the Mantel-Haenszel method for dichotomous data, assigning the weight of each study based on its variance. All meta-analyses were conducted using a random-effects model to account for expected clinical and methodological heterogeneity across studies, including differences in study design, populations, and outcome measurements. A direct comparison between the two techniques was conducted by assessing studies that reported outcomes of both treatments (two-arm analysis). The heterogeneity of effect size estimates across the studies was quantified using the
Results
The literature search of the electronic databases identified a total of 1614 studies. After duplicates were removed, 1434 articles were screened, and 117 remained for full-text review. Those articles were then assessed for eligibility using specified inclusion and exclusion criteria. Of those articles, 16 studies satisfied the eligibility criteria and were included (PRISMA flow chart Figure 1(a)). Kessing et al. 35 and Gislason et al. 36 were found and added through manual search after the literature search was completed. The baseline characteristics of the studies included are comprehensively described in Table 1.
Baseline characteristics of the studies included in the meta-analysis, including sample size, study design, and population characteristics.
NR, not reported; SD, standard deviation.
Baseline clinical characteristics
Among the 18 studies, a total of 476,693 patients were included, of which 41,600 were taking lithium, and 435,188 were not.
Mean eGFR and annual decline in eGFR outcomes
Mean eGFR was compared between the lithium and non-lithium groups (Figure 2(a)) and showed less favorable outcomes in the lithium group compared to the non-lithium group (
Comparison of estimated glomerular filtration rate (eGFR) between lithium-exposed and non-lithium groups. (a) Forest plot of mean eGFR (ml/min/1.73 m²) in lithium users versus non-lithium controls. (b) Forest plot of annual decline in eGFR (ml/min/1.73 m² per year) in lithium users versus non-lithium controls.
Serum creatinine outcomes
Serum creatinine concentration (mg/dl) outcome was found to be higher in the lithium compared to the non-lithium groups (
Comparison of serum creatinine levels (mg/dl) between lithium-exposed and non-lithium groups.
New or progressing chronic kidney disease
New or progressing CKD events were found to be comparable between the lithium and non-lithium groups (
Incidence of new or progressive chronic kidney disease (CKD) in lithium-exposed versus non-lithium groups.
Risk of bias
Results of the quality assessment of all included studies are shown in Figure 1(b). Out of the studies assessed, one study had a low risk of bias, 42 fourteen studies were of high risk26,27,29,35,36,40,41,43–49 and three studies were of very high risk.37–39 Those studies were found to be of high risk or very high concern due to not properly accounting for confounders and covariates, selection bias due to including hospital-only patients, and bias due to post-exposure interventions, given that patients on lithium were more likely to get regular renal function measurements. Despite these concerns, all studies sufficiently reported the relevant outcome domains. The risk of bias assessments informed our interpretation of the pooled findings.
Discussion
Impaired kidney function with long-term lithium use in populations with BD remains a contested subject despite extensive research. The main aim of this meta-analysis was to better understand kidney function in BD patients on lithium compared to patients on alternative pharmacotherapies or not on any current bipolar medical therapy. In our meta-analysis, six studies, Coskunol et al.,
39
Dastych et al.,
40
Hayes et al.,
42
Jonczyk-Potoczna et al.,
44
Turan et al.,
49
and Kuruvilla et al.,
45
were found to include only patients with BD. Dastych et al.
40
found no significant differences in creatinine or eGFR between long-term lithium-treated patients and BD patients never on lithium, treated with other mood stabilizers, but reported decreased urine osmolarity (mean ± SD, 405 ± 164 vs 667 ± 174 mmol/kg) and urine-to-serum osmolality ratio (1.35 ± 0.61 vs 2.25 ± 0.96) in the lithium-treated group. Jonczyk-Potoczna et al.,
41
on the other hand, further analyzed lithium-treated BD patients by separating them into patients with renal macrocysts and patients without. Lithium-treated patients with renal macrocysts had overall poorer renal function with higher serum creatinine levels and lower eGFR rates compared to lithium-treated patients without macrocysts. In their study, renal macrocysts were reported in 22% of patients on lithium versus 16% not on lithium, with a non-statistically significant difference. These findings might suggest that certain predisposing factors, such as renal macrocysts, can increase nephrotoxicity susceptibility in a subset of BD patients treated with lithium long-term. In an older study with a higher risk of bias, Kuruvilla et al.,
45
reported normal eGFR and serum creatinine in both test (
Our results show a significant decrease in mean eGFR and higher levels of serum creatinine in the lithium versus the non-lithium groups. This agrees with the findings of a multicenter prospective study evaluating kidney function with long-term lithium exposure in 312 BD patients. 50 However, in a 2012 meta-analysis by McKnight et al. 51 on the lithium toxicity profile, it was found that eGFR decline with lithium use was not statistically significant. The discrepancy in results could be reconciled by Turan et al.’s results. 49 Turan et al. 49 compared the renal function of long-term lithium-use patients and short-term lithium-use patients to the control groups. This meta-analysis included the long-term exposure group’s outcomes, which showed significant differences from the control group, while McKnight et al. 51 included short-term exposure group outcomes. This further emphasizes the relationship between the length of exposure to lithium and renal function decline. Indeed, Tondo et al.’s 50 results showed an added 30% decline in eGFR in people on long-term lithium treatment compared to the decline associated with aging alone. This decline was found to be associated with several risk factors, namely older age, comorbidities, lower starting eGFR, longer lithium treatment duration, and higher serum lithium level. 50 The proportion of this 30% decline in eGFR that is attributable to the aforementioned risk factors is unclear, making this result more likely reflective of a lithium-associated nephrotoxicity rather than a direct consequence of lithium exposure, as would be the case in lithium-induced nephrotoxicity. Regular monitoring of renal function and appropriate management of nephrotoxicity risk factors are, therefore, essential aspects of clinical care for patients receiving lithium therapy.
Interestingly, when looking at the outcomes of large-sample cohort studies, Bosi et al.
26
(
New or progressing CKD events were found to be comparable between the lithium groups and the control groups. A leave-one-out sensitivity analysis was performed to assess the outcome’s odds ratio (OR), excluding outlier Bosi et al. results. This showed a statistically significant OR of developing or progressing CKD in lithium-treated patients (OR = 3.40, 95% CI: 2.83, 4.09,
The mechanistic and histological alterations by which lithium exerts its toxicity on the kidneys have been extensively studied. 54 Lithium affects the immune-inflammatory system by increasing the leukocyte count.55,56 Data regarding the proinflammatory effects of lithium in mice seem contradictory. Indeed, a study showed that lithium leads to the amplification of proinflammatory mediators, ultimately leading to pyroptotic renal cell death. 57 In contrast, another article shows no changes to proinflammatory responses despite also increasing kidney cell apoptosis. 58 Importantly, these contrasting findings may be reconciled by the fact that the mice from Jing et al had blood lithium concentrations close to 3 mmol/l, whereas those from Baranovskaya were maintained within the 0.7–1.5 mmol/l serum concentration range, below the toxicity threshold in psychiatric patients. 58 In addition, lithium-subjected rats showed modified renal architecture, dysregulation in the expression of apoptotic and inflammatory proteins, and increased Malondialdehyde (MDA) levels, indicating increased oxidative stress. 59 Similarly, transforming growth factor 2 (TGF-β2), a central player in renal fibrosis, was overexpressed in the cortex and medulla of lithium-fed rats’ kidneys. 60 Furthermore, studies on the effect of lithium on human and mouse kidneys revealed renal damage, including tubular atrophy, chronic interstitial fibrosis, and glomerulosclerosis (the latter being observed in some cases).61–63 Long-term use of lithium increases the risk of renal tubular dysfunction, which in later stages may progress to structural alteration.63,64
Across the assessed studies in this analysis, measures of renal function show unfavorable or similar values when comparing lithium-taking groups with controls. Indeed, parameters including eGFR and creatinine clearance suggest some nephrotoxic effects of lithium. However, the studies did not adequately adjust for covariates or match for covariates at baseline. It is therefore important to consider the variability in patients’ baseline renal function and the impact of psychiatric comorbidities when evaluating the long-term effects of lithium. Patients with BD are at a higher risk of cardiovascular diseases, such as hypertension and heart failure, which are independent risk factors for the development of renal insufficiency. Some of the studies have reported a higher baseline creatinine among patients who developed chronic kidney disease on lithium, highlighting the propensity to develop renal insufficiency, rather than the effect of lithium. This observation thus makes any documented nephrotoxic effects within the studied cohorts more appropriately classifiable as lithium-associated and not lithium-induced. Indeed, potential confounding variables that may disproportionately be represented in lithium-taking patients can cloud the nature of the relationship linking this drug to the measured adverse effect of interest. In contrast, assessments of albuminuria and CKD incidence or progression do not support this outcome.
In addition, this meta-analysis highlights that the necessity for continuous monitoring of renal function is not an end, but rather a critical component of a proactive risk mitigation strategy. In fact, regular assessment of renal function allows clinicians to make informed decisions, such as adjusting the lithium dose to a lower but still therapeutic level or considering the gradual cessation of the medication in cases of significant or progressive renal impairment. In taking these clinical decisions, clinicians must evidently weigh the well-established benefits of lithium, particularly its antisuicidal properties, against its cessation. Moreover, clinicians should take into consideration that abrupt discontinuation of lithium may induce a rapid relapse or withdrawal phenomenon that can confound the true long-term prophylactic benefits of the medication. Indeed, Baldessarini et al. 65 provide evidence for this claim, as their study shows that among patients who discontinued lithium, suicide rates rose 20-fold in the first year off the medication compared to the period when they were on it. This same study also found that affective illness recurred in 67% of patients in the year following discontinuation. 65 These findings suggest that the high rates of relapse are not solely due to the re-emergence of the underlying illness but are significantly influenced by a withdrawal effect. Therefore, discussions about discontinuing lithium should always involve a gradual tapering schedule to mitigate these withdrawal manifestations. The work by Baldessarini et al. also shows that the early morbidity was 2.5-fold lower and suicidal risk was 2.0-fold lower after a slow (15–30 days vs rapid 1–14 days) discontinuation. This finding underscores the importance of a carefully managed withdrawal process.
Limitations
As with all meta-analyses, this study has several limitations. Patients with BD constitute only a small subset of the population studied in most of the literature investigating lithium nephrotoxicity. The non-lithium population included in comparative studies is highly variable, with some papers including the general population as a control, while others included patients with psychiatric diagnoses not on lithium or on antipsychotic or antiepileptic medications. These limitations posed a significant challenge to our ability to draw any meaningful conclusions about the nephrotoxic effect of lithium in BD patients and their baseline kidney function when not on any bipolar medications or when on alternative therapies.
Except for Fransson et al., Rej et al., Hayes et al., and Minay et al.,27,46,47,66 most of the included studies rely on relatively small sample sizes to formulate their results. Furthermore, publications such as Coppen et al., Gelenberg et al., Hullin et al., Kuruvilla et al., and Wahlin et al.29,41,43,45,38 are rather dated, and this could compromise the robustness of the included data. Indeed, Kuruvilla et al. 45 included only 64 subjects, which is a relatively small sample size for a long-term study on a chronic condition. Hulin et al. 43 also used a very small sample of only 30 lithium patients for the main comparative analysis. Moreover, Hullin et al. 43 used a cross-sectional design where patients were admitted overnight for a single, short period of investigation. Coppen et al. 38 had a larger sample of 101 patients on lithium but failed to account for the long-term use of other drugs, which could have independently affected renal function. Wahlin et al. 29 compared only 25 patients with affective disorders who had never received lithium against healthy controls. These factors, which are better controlled for in more recent prospective studies, are critical for a precise interpretation of lithium’s long-term effects on kidney function. By contrast, more recent studies with larger cohorts and more robust designs offer a more reliable basis for drawing conclusions about lithium’s nephrotoxicity.
Conclusion
Care should be taken to adequately monitor renal function in lithium-receiving patients by assessing baseline values and comparing them with periodic follow-up levels. Either way, this apparent side effect of lithium must not be overlooked when discussing possible therapies to allow for educated patient-driven decision-making in conjunction with caregivers. The recent studies that use more balanced designs, adjusting for various medications and employing novel study approaches, such as active-comparator designs, offer a more balanced perspective on kidney-related side effects of lithium. According to recent findings, these effects seem comparable to those of valproate. Future research should focus on comparing the risk of renal insufficiency between patients on lithium and those on second-generation atypical antipsychotics, as these are the most commonly prescribed medications for BD. Second-generation antipsychotics are associated with a higher risk of cardiometabolic diseases, which may predispose patients to chronic kidney disease.
Furthermore, effective treatment options tailored to mitigate potential lithium nephrotoxicity risks should be established to manage BD, and this is vital for patients who already suffer from decreased renal function. This paper highlights the need for additional prospective and randomized controlled trial studies with long-term follow-ups. Identifying individuals at higher risk of developing chronic kidney disease could be crucial. Managing the medical comorbidities associated with these conditions may help slow the progression of renal insufficiency or the development of chronic kidney disease in patients taking lithium. While conducting long-term randomized controlled trials comparing the risk of renal insufficiency between lithium and other pharmacotherapies for mood disorders would be challenging, studies such as inhalation trials may offer valuable insights to help address this complex issue. Identifying potential biomarkers, genetic drivers, or environmental factors behind increased vulnerability to lithium-induced nephrotoxicity could also limit patient risks by reframing criteria that make this treatment favorable on a personalized, case-by-case basis.
Supplemental Material
sj-docx-1-tpp-10.1177_20451253261419633 – Supplemental material for Lithium nephrotoxicity: a systematic review and meta-analysis of lithium versus non-lithium control studies in patients with affective disorders
Supplemental material, sj-docx-1-tpp-10.1177_20451253261419633 for Lithium nephrotoxicity: a systematic review and meta-analysis of lithium versus non-lithium control studies in patients with affective disorders by Marie Michele Macaron, Mireilla Abou Rjeily, Raphael Macaron, Asmaa Yehia, Balwinder Singh, Mark A. Frye and Osama A. Abulseoud in Therapeutic Advances in Psychopharmacology
Supplemental Material
sj-docx-2-tpp-10.1177_20451253261419633 – Supplemental material for Lithium nephrotoxicity: a systematic review and meta-analysis of lithium versus non-lithium control studies in patients with affective disorders
Supplemental material, sj-docx-2-tpp-10.1177_20451253261419633 for Lithium nephrotoxicity: a systematic review and meta-analysis of lithium versus non-lithium control studies in patients with affective disorders by Marie Michele Macaron, Mireilla Abou Rjeily, Raphael Macaron, Asmaa Yehia, Balwinder Singh, Mark A. Frye and Osama A. Abulseoud in Therapeutic Advances in Psychopharmacology
Footnotes
References
Supplementary Material
Please find the following supplemental material available below.
For Open Access articles published under a Creative Commons License, all supplemental material carries the same license as the article it is associated with.
For non-Open Access articles published, all supplemental material carries a non-exclusive license, and permission requests for re-use of supplemental material or any part of supplemental material shall be sent directly to the copyright owner as specified in the copyright notice associated with the article.
