Potential nephrotoxicity from long-term lithium use has been well documented for many decades, with the first paper demonstrating structural renal damage, primarily in the form of interstitial nephritis (with scarring of the interstitium, tubular destruction and relative preservation of glomeruli), published in 1977.¹ The trajectory of renal function in patients who have had lithium treatment discontinued due to progressive renal impairment (previously measured by serum creatinine, more recently measured by eGFR [estimated glomerular filtration rate]) is still not clear. Studies have shown that, after lithium discontinuation, renal function either stabilizes or improves in many patients while other patients, especially those with worse renal function at the time of lithium discontinuation, show further deterioration in renal function.²

Among the variables that may explain these discrepant results are: (1) the effect of comorbid medical disorders that independently adversely affect renal function such as diabetes or hypertension; (2) the extent of renal damage at the time of lithium discontinuation. As an example, in one recent study, individuals most likely to show progressive deterioration of renal function after lithium discontinuation were those whose eGFR was 32 mL/min or less.² This finding is consistent with previous studies showing similar results. (3) A final variable, which may be the most important of all, is the length of follow-up after lithium discontinuation. Many studies in this area have followed patients for 1–5 years. However, renal function normally and inevitably deteriorates with age and it may take decades for mild to moderate chronic kidney disease (CKD) (e.g. stage 3, with an eGFR<60) to progress to end-stage renal disease (ESRD).

Beyond the frustration of conflicting data, the key issue is that of balancing the risk of treating vs. the risk of not treating, that is removing the treatment that is beneficial. For lithium, the risk of continuing lithium treatment in the face of progressive deterioration in renal function is that of inexorable renal damage. The risk of not treating reflects the frequent observation that, for many patients, treatment with lithium can be either extraordinarily effective or even life-saving given the morbidity of recurrent bipolar episodes. Consistent with this, one recent study demonstrated that patients who discontinued lithium due to CKD stage 3 had more mood episodes, and more psychotropic medication trials compared to those who continued lithium treatment.³

The UCLA Mood Disorders Clinic has been providing longitudinal care to patients with unipolar and bipolar disorder for 45 years. Therefore, we have followed a number of patients for many decades. To illustrate the importance and effect of long-term observation, four brief clinical vignettes of patients with bipolar I disorder treated with lithium are presented, demonstrating both the potential for progression to ESRD in three patients, the toxic effects of dialysis in one of these patients and mood destabilization from discontinuing lithium in another patient.

Mr. A, previously reported on 30 years ago as Mr. C in Gitlin, 1993,⁴ with bipolar I disorder, was treated with lithium. He developed a rising serum creatinine, which was 2.1 mg/100 mL at the time of lithium discontinuation after 12 years of treatment. After lithium discontinuation, he developed a treatment refractory depression. His renal function deteriorated thereafter and he was eventually placed on dialysis treatment. When it was clear that he was inevitably progressing to ESRD and dialysis or renal transplantation, his lithium was restarted with resolution of his depression. However, while on haemodialysis, he continued to have recurrent mood episodes despite the addition of other mood stabilizers to his lithium treatment. The patient refused renal transplantation due to his fear of the requisite corticosteroid treatment post-transplant and its potential to precipitate another manic episode. (He had experienced a hospitalized manic episode in the past associated with corticosteroid use). He eventually developed dialysis-related dementia and died at age 64 of an arrhythmia due to metabolic imbalance related to his dialysis treatment.

Ms. B (also previously reported on in Gitlin, 1993⁴ after 6 years of lithium treatment developed a serum creatinine of 2.5 mg/100 mL. After valproate plus bupropion were added, lithium was tapered and discontinued. Over the next 5 years, her serum creatinine decreased to 1.7 mg/100 mL. She did very well for decades on valproate plus bupropion. However, over the next 30+ years, off lithium, her renal function slowly and progressively deteriorated to a current eGFR = 13, meeting diagnostic criteria for ESRD (eGFR<15). Now aged 65 years old, she is preparing for dialysis or renal transplantation.

Ms. C, now 51 years old, took lithium with excellent effect for 6 years. In 2000, her lithium was discontinued due to a serum creatinine of 1.8–2.0 ng/100 mL. After 2 years of mood instability, she was eventually well-stabilized on oxcarbazepine plus lamotrigine. Over the next 23 years, her renal function gradually deteriorated. Currently, her eGFR <15, meeting criteria for ESRD and she is preparing for renal transplantation.

Ms. D, after very effective long-term treatment with lithium, was evaluated by a nephrologist who recommended discontinuing lithium because of a progressive decrease in her eGFR. Lithium was discontinued when her eGFR was 38 (CKD stage 3b). Despite many other therapies, including other mood stabilizers and ECT, she died by suicide within 1 year of discontinuing lithium.

These four cases illustrate a set of difficult clinical decisions regarding long-term lithium use. The absolute risk for ESRD in lithium-treated is very low, even if the relative risk is substantial. Recent estimates of ESRD associated with lithium range between 0.2 and 0.7%, which is an almost eightfold risk compared to the general population.⁵ But, clinically, how does one weigh the relative risk of end-stage renal disease compared to the potential for a less well-treated bipolar I disorder? Ms. B and Ms. C were both successfully treated with other mood stabilizers after lithium discontinuation. Ms. D was clearly not. Mr. A continued to have recurrent mood cycles both on and off lithium. Despite the cycling while on lithium, he felt that his mood swings, while present were milder when taking lithium.

Additionally, how does one accurately and succinctly describe the risk/benefit issues about lithium with bipolar patients as part of informed consent? Does one acknowledge the rare but potentially increased risk for ESRD decades in the future at the time of initiating lithium treatment? Does one discuss this only if and at the time of a progressive decline in eGFR? Finally, at what level of renal function should lithium discontinuation be considered and discussed? The literature suggests that eGFR = 40 mL/min or less is associated with greater risk of further renal function deterioration. Should the discussion occur then or earlier, if/when the patient develops CKD stage 3 (eGFR <60 mL/min)?

As a field, we must grapple with these thorny dilemmas, which have no clear answers but real clinical consequences. Most importantly, however, we should at least remind ourselves that short-term (measured in a few years) observations may be reassuring but may underestimate very long-term outcomes that are seen only after decades of observation.

几十年来，长期使用锂的潜在肾毒性已得到充分记录，第一篇论文证明了结构性肾损伤，主要以间质性肾炎的形式（伴有间质疤痕、肾小管破坏和肾小球相对保留），发表于1977. ¹由于进行性肾功能损害（之前通过血清肌酐测量，最近通过 eGFR [估计肾小球滤过率] 测量）而停止锂治疗的患者的肾功能轨迹仍不清楚。研究表明，停药后，许多患者的肾功能稳定或改善，而其他患者，特别是停药时肾功能较差的患者，肾功能进一步恶化。²

可以解释这些差异结果的变量包括：（1）独立对肾功能产生不利影响的共存疾病的影响，例如糖尿病或高血压；(2) 停用锂剂时肾损害的程度。例如，在最近的一项研究中，停用锂剂后最有可能出现肾功能进行性恶化的个体是 eGFR 为 32 mL/min 或更低的个体。²这一发现与之前显示类似结果的研究一致。(3) 最后一个变量，可能是最重要的，是停用锂盐后的随访时间长度。该领域的许多研究都对患者进行了 1-5 年的跟踪研究。然而，肾功能通常不可避免地会随着年龄的增长而恶化，轻度至中度慢性肾病（CKD）（例如3期，eGFR<60）可能需要几十年的时间才能进展为终末期肾病（ESRD）。

除了相互矛盾的数据带来的挫败感之外，关键问题是平衡治疗风险与不治疗风险，即取消有益的治疗。对于锂盐来说，在肾功能逐渐恶化的情况下继续锂盐治疗的风险是不可避免的肾损伤。不治疗的风险反映了经常观察到的情况，即对于许多患者来说，考虑到双相情感障碍反复发作的发病率，锂盐治疗可能非常有效，甚至可以挽救生命。与此相一致的是，最近的一项研究表明，与继续锂盐治疗的患者相比，因 CKD 3 期而停止锂盐治疗的患者有更多的情绪发作和更多的精神药物试验。³

45 年来，加州大学洛杉矶分校情绪障碍诊所一直为单相和双相情感障碍患者提供纵向护理。因此，我们对许多患者进行了数十年的跟踪研究。为了说明长期观察的重要性和效果，提供了四例接受锂治疗的 I 型双相情感障碍患者的简短临床案例，证明了三名患者进展为 ESRD 的可能性，以及其中一名患者透析的毒性作用另一位患者因停止服用锂而情绪不稳定。

A 先生，30 年前曾在 Gitlin，1993 年报道为 C 先生，⁴患有 I 型双相情感障碍，曾接受锂治疗。他的血清肌酐升高，在治疗 12 年后停止锂盐治疗时，血清肌酐为 2.1 mg/100 mL。停用锂剂后，他出现了难治性抑郁症。此后，他的肾功能恶化，最终接受透析治疗。当很明显他不可避免地会进展为终末期肾病（ESRD）和透析或肾移植时，他重新开始服用锂盐，抑郁症得到缓解。然而，在血液透析期间，尽管在锂治疗中添加了其他情绪稳定剂，但他的情绪仍然反复发作。该患者拒绝肾移植，因为他担心移植后必要的皮质类固醇治疗及其可能引发另一次躁狂发作。（他过去曾经历过与使用皮质类固醇相关的住院躁狂发作）。他最终患上了与透析相关的痴呆症，并因透析治疗引起的代谢失衡而死于心律失常，享年 64 岁。

B 女士（Gitlin，1993 年也曾报道过）⁴经过 6 年锂治疗后，血清肌酐为 2.5 mg/100 mL。添加丙戊酸加安非他酮后，锂逐渐减量并停用。在接下来的 5 年里，她的病情血清肌酐降至 1.7 mg/100 mL。几十年来，她在丙戊酸钠加安非他酮治疗下表现良好。然而，在接下来的 30 多年里，在停用锂剂后，她的肾功能缓慢且逐渐恶化，目前的 eGFR = 13，符合诊断标准患有终末期肾病（eGFR<15）。现年65岁，正在准备透析或肾移植。

C女士现年51岁，服用锂剂6年，效果极佳。2000 年，由于血清肌酐为 1.8-2.0 ng/100 mL，她停止服用锂盐。经过两年的情绪不稳定后，她最终在奥卡西平加拉莫三嗪的治疗下病情稳定下来。接下来的23年里，她的肾功能逐渐恶化。目前，她的eGFR<15，符合ESRD标准，正在准备肾移植。

D 女士经过非常有效的长期锂盐治疗后，肾病专家对她进行了评估，由于她的 eGFR 逐渐下降，建议停止使用锂盐。当她的 eGFR 为 38（CKD 3b 期）时，停止服用锂盐。尽管有许多其他治疗方法，包括其他情绪稳定剂和 ECT，但她在停止服用锂剂后 1 年内自杀身亡。

这四个案例说明了有关长期使用锂的一系列困难的临床决策。即使相对风险很大，锂处理后的 ESRD 绝对风险也非常低。最近估计与锂相关的 ESRD 范围在 0.2% 至 0.7% 之间，与一般人群相比，该风险几乎是普通人群的八倍。⁵但是，在临床上，与治疗较差的 I 型双相情感障碍的可能性相比，如何权衡终末期肾病的相对风险？B 女士和 C 女士在停用锂剂后均使用其他情绪稳定剂成功治疗。D女士显然不是。A 先生在使用或不使用锂盐时仍然会出现反复的情绪周期。尽管服用锂盐期间骑自行车，但他觉得服用锂盐时他的情绪波动较温和。

此外，作为知情同意的一部分，如何准确、简洁地描述双相情感障碍患者使用锂的风险/效益问题？在开始锂治疗时，人们是否承认未来几十年出现罕见但可能增加的 ESRD 风险？是否只有在 eGFR 逐渐下降时才讨论这一问题？最后，肾功能达到什么水平时应考虑和讨论停用锂盐？文献表明，eGFR = 40 mL/min 或更低与肾功能进一步恶化的风险更大相关。如果/当患者发展为 CKD 3 期（eGFR <60 mL/min）时，讨论应该在当时还是更早进行？

作为一个领域，我们必须解决这些棘手的困境，这些困境没有明确的答案，但有真正的临床后果。然而，最重要的是，我们至少应该提醒自己，短期（几年内衡量）的观察可能令人放心，但可能会低估只有经过几十年的观察才能看到的长期结果。