Lithium Nephrotoxicity: a Progressive Combined Glomerular and Tubulointerstitial Nephropathy

Home , Nephrotoxicity

J Am Soc Nephrol 11: 1439–1448, 2000 Lithium Nephrotoxicity: A Progressive Combined Glomerular and Tubulointerstitial Nephropathy

GLEN S. MARKOWITZ,* JAI RADHAKRISHNAN,† NEERAJA KAMBHAM,* ANTHONY M. VALERI,† WILLIAM H. HINES,‡ and VIVETTE D. D’AGATI* Departments of *Pathology and †Medicine, Columbia Presbyterian Medical Center, New York, New York; and ‡Department of Medicine, Stamford Hospital, Stamford, Connecticut.

Abstract. This study examines the clinical features, pathologic surprisingly high prevalence of focal segmental glomeruloscle- findings, and outcome of 24 patients with biopsy-proven lith- rosis (50%) and global glomerulosclerosis (100%), sometimes ium toxicity. The patient population was 50% male, 87.5% of equivalent severity to the chronic tubulointerstitial disease. Caucasian, and had a mean age of 42.5 yr (range, 26 to 57). The significant degree of foot process effacement (mean 34%, Mean duration of lithium therapy for bipolar disorder was 13.6 five of 14 cases with Ͼ50%) suggests a potential direct glo- yr (range, 2 to 25). All patients were biopsied for renal insuf- merular toxicity. Focal segmental glomerulosclerosis corre- ficiency (mean serum creatinine 2.8 mg/dl; range, 1.3 to 8.0), lated with proteinuria Ͼ1.0 g/d (P ϭ 0.0014, Fisher exact test). with associated proteinuria Ͼ1.0 g/d in 41.7%. Nephrotic Despite discontinuation of lithium, seven of nine patients with proteinuria (Ͼ3.0 g/d) was present in 25%. Other features initial serum creatinine values Ͼ2.5 mg/dl progressed to end- included nephrogenic diabetes insipidus in 87% and hyperten- stage renal disease (ESRD). Only three patients, all with initial sion in 33.3%. Renal biopsy revealed a chronic tubulointersti- serum creatinine Ͻ2.1 mg/dl, had subsequent improvement in tial nephropathy in 100%, with associated cortical and medul- renal function. By Kaplan–Meier survival analysis, the only lary tubular cysts (62.5%) or dilatation (33.3%). All of the significant predictor of progression to ESRD was serum cre- renal cysts stained for epithelial membrane antigen, while atinine Ͼ2.5 mg/dl at biopsy (P ϭ 0.008). In conclusion, 51.4% stained with lectin Arachis hypogaea, and only 3.8% lithium nephrotoxicity primarily targets distal and collecting stained with Tetragonolobus purpureas, indicating they origi- tubules, with a higher incidence of proteinuria and associated nated from distal and collecting tubules. The degree of tubular glomerular pathology than recognized previously. Renal dys- atrophy and interstitial fibrosis was graded as severe in 58.3%, function is often irreversible despite lithium withdrawal, and moderate in 37.5%, and mild in 4.2% of cases. There was a early detection is essential to prevent progression to ESRD.

Lithium is a therapeutic agent currently in widespread use for the water channel aquaporin-2, expressed on the apical plasma treatment of bipolar disorder. Although this agent is highly effec- membrane of principal cells of the collecting duct (2). Acute tive in reducing symptoms of manic depression, a frequent side lithium intoxication due to lithium overdose may be seen in effect is renal toxicity. A major biochemical action of lithium in patients on chronic lithium therapy or in lithium-naı¨ve indi- the kidney is competition with magnesium, thereby inhibiting viduals. Acute effects include a change in mental status, acute magnesium-dependent G proteins that activate vasopressin-sensi- renal failure, and volume depletion. The primary management tive adenylyl cyclase (1). Lithium nephrotoxicity can be divided of acute lithium intoxication is hemodialysis (1). into three main categories: nephrogenic diabetes insipidus, acute The predominant form of chronic renal disease associated with intoxication, and chronic renal disease. lithium therapy is a chronic tubulointerstitial nephropathy (CTIN) Nephrogenic diabetes insipidus (NDI) is the most common that is heralded by the insidious development of renal insuffi- renal side effect of lithium therapy. Patients present with ciency, with little or no proteinuria, often in the setting of chronic polyuria and polydipsia due to a urinary concentrating defect NDI. Biopsy findings in patients with lithium-induced CTIN that can lead to significant volume depletion. Experimental include tubular atrophy and interstitial fibrosis, typically out of studies have shown that the development of NDI involves proportion to the degree of glomerulosclerosis or vascular disease lithium-induced downregulation of the vasopressin-regulated (3–6). The majority of studies have shown infrequent and relatively mild renal insufficiency attributable to lithium therapy (7). Only one case of end-stage renal disease (ESRD) secondary to Received November 4, 1999. Accepted December 22, 1999. lithium-associated CTIN has been reported (8). Correspondence to Dr. Vivette D. D’Agati, Department of Pathology, Colum- Much less has been written about the potential glomerular bia Presbyterian Medical Center, 630 West 168th Street, VC14-224, New toxicity of lithium. There are many documented cases of min- York, NY 10032. Phone: 212-305-7460; Fax: 212-342-5380; E-mail: [email protected] imal change disease occurring in association with lithium ther- 1046-6673/1108-1439 apy (9,10). Some cases have resolved following lithium with- Journal of the American Society of Nephrology drawal and recurred after its reintroduction, providing strong Copyright © 2000 by the American Society of Nephrology evidence of an etiologic relationship (9,10). Only three cases of 1440 Journal of the American Society of Nephrology J Am Soc Nephrol 11: 1439–1448, 2000 focal segmental glomerulosclerosis (FSGS) in patients receiv- been treated with lithium in standard doses for a mean of 13.6 ing lithium carbonate have been described (11). yr (range, 2 to 25 yr). All patients presented with renal insufficiency (Table 3). Materials and Methods The mean initial serum creatinine was 2.8 mg/dl with a range of 1.3 to 8.0 mg/dl. The patient who presented with a serum All renal biopsies processed from 1986 to 1999 in the Renal Pathology creatinine of 8.0 mg/dl required emergent hemodialysis and Laboratory at Columbia Presbyterian Medical Center were reviewed for a history of lithium therapy. Twenty-four biopsies from patients treated remained dialysis-dependent. Only two patients were known to with lithium were identified from among 6514 native kidney biopsies have previous episodes of acute lithium intoxication (patients 7 (0.37%). All cases were processed for light microscopy, immunofluores- and 18). Proteinuria was a common presenting feature: 10 cence, and electron microscopy according to standard techniques. The patients (41.7%) had a 24-h protein measurement of Ͼ1.0 g/d, tissue available for electron microscopy included glomeruli in 15 cases and six patients (25%) had nephrotic-range proteinuria. Three and tubules in 24 cases. Each case was reviewed independently by two of these six patients (patients 1, 2, and 19) met two of the three renal pathologists. Tubular atrophy and interstitial fibrosis were graded additional criteria for the diagnosis of nephrotic syndrome on a scale of mild, moderate, and severe, corresponding to involvement (hypoalbuminemia, hypercholesterolemia, and peripheral ede- Ͼ of 0 to 25, 26 to 50, and 50% of the cortex sampled, respectively. ma). Hematuria and leukocyturia were present in two and three Tubular cysts were defined as tubules with a diameter of at least five patients, respectively. Interestingly, ultrastructural analysis re- times that of normal tubules. Tubular profiles with a diameter 2 to 4 times vealed underlying thin basement membrane disease in one of that of normal tubules were considered to have tubular dilatation, likely representing incipient cysts. The percentages of globally and segmentally the three patients with hematuria (patient 10). Clinical evi- sclerotic glomeruli were recorded. dence of NDI was present in 87% of patients. Patient charts were reviewed retrospectively for presenting symp- All 24 renal biopsy samples were believed to be adequate for toms, laboratory data at presentation, and follow-up (up to 11 yr). diagnosis, with a mean glomerular count of 29.3 (range, 4 to Nephrotic-range proteinuria was defined as a 24-h urine collection of Ͼ100; median 20). The predominant pathologic finding ob- Ͼ3.0 g/d. Hypoalbuminemia was defined as serum albumin Ͻ3.3 served in all 24 of the biopsy samples was CTIN, characterized g/dl, and hypercholesterolemia was defined as serum cholesterol by tubular atrophy and interstitial fibrosis, out of proportion to Ͼ200 mg/dl. Nephrogenic diabetes insipidus was diagnosed based on the extent of glomerular or vascular disease (Table 4, Figure 1, clinical criteria of polyuria, polydipsia, and consistently low urine- A and B). Tubular atrophy and interstitial fibrosis were graded Ͻ specific gravities ( 1.010). as mild in one patient, moderate in nine patients, and severe in Renal cysts remained in the tissue block from eight of the biopsy 14 patients. In all patients, the degree of tubular atrophy and samples. To determine the nephron segments forming cysts, these cases were stained with lectins Tetragonolobus purpureas (TP) for interstitial fibrosis was either equal to (21.7%) or greater than proximal tubules, and Arachis hypogaea (AH) for collecting tubules, (78.3%) the degree of arteriosclerosis. Tubulointerstitial dis- and with antibody to epithelial membrane antigen (EMA) for distal ease tended to be patchy, with geographic zonation, particu- and collecting tubules, according to standard techniques (12). Briefly, larly in the less advanced cases. Although a sparse predomi- deparaffinized tissue sections were microwave-digested in ethylenedi- nantly lymphocytic interstitial infiltrate was often confined to aminetetra-acetic acid (pH 7.4), quenched in 1% H2O2 in ethanol, and areas of interstitial fibrosis, there was no associated tubulitis. then overlaid with peroxidase-conjugated lectins TP and AH (Sigma Tubular cysts, a common finding in patients receiving lithium Chemical Co., St. Louis, MO) at 1:6 and 1:100 dilutions for 60 min at (13), were observed in 62.5% of the renal biopsies (Figure 1C). room temperature, respectively. Color reactions were developed with Cysts were most frequently identified in the zones of tubular diaminobenzidine (DAB). Immunohistochemical staining for EMA atrophy and interstitial fibrosis, both in the cortex and the medulla. (Dako, Carpinteria, CA) was performed at a dilution of 1:50 with They tended to be sparse and did not exceed 1 to 2 mm in microwave digestion, ABC technique, and DAB color reagent. diameter. A pericystic cuff of fibrosis was sometimes present, with associated thickening of the cystic tubular basement mem- Statistical Analyses brane. The cystic epithelial cells displayed cytologic features of Statistical analyses were performed using ESRD as the end point distal or collecting tubular epithelium, consisting of cuboidal cells and included univariate analysis with Pearson ␹2 test, Fisher exact with relatively clear cytoplasm and no identifiable brush border. test, independent samples t test, and Kaplan–Meier survival estimates, The cytoplasmic ballooning and periodic acid-Schiff-positive as appropriate, and multivariate analysis by logistic regression and the granules described by Burrows et al. (14) were rarely observed. Cox proportional hazards regression analysis. Statistical significance Mild epithelial hyperplasia of the cyst lining cells was identified was assumed at P Ͻ 0.05. rarely. Dilated tubules displayed similar cytologic features but usually lacked the pericystic fibrous cuff (Figure 1C). Results Immunohistochemical staining for segment-specific markers The clinical presentations, biopsy findings, and outcomes of including EMA and lectin staining for TP and AH were per- each patient are listed in Table 1 and summarized in Tables 2 formed on the eight biopsies in which cystic tissue remained in through 5. The patient population included 24 patients, with a the paraffin block (Table 5, Figure 2). Two cases were open male-to-female ratio of 1.0 (Table 2). The majority of the renal biopsies with abundant cystic tissue (patients 14 and 24), patients were Caucasian (87.5%), and all patients had been while the other six cases displayed up to three tubular cysts. A diagnosed with bipolar disorder. At the time of biopsy, one total of 105 cysts was identified and all stained positively with patient was diabetic and eight were hypertensive. Patients had EMA. In contrast, 51.4% stained with AH and 3.8% stained mScNprl1:13–48 2000 1439–1448, 11: Nephrol Soc Am J

Table 1. Clinical presentation, biopsy findings, and follow-up in patients who received lithium therapya

Patient Variable 1 2 3 4 5 6 7 8 9 101112131415161718192021222324

Age 51 37 57 56 26 34 40 35 51 44 36 55 27 50 55 45 33 40 32 38 36 41 44 56 Race W W W W AA W W W W W W W W W W H W W W W W AA W W Gender M F F M M F F F F F F M F M M M F M M F M M F M Presentation RI/Pr RI/Pr RI RI/Pr RI/Pr RI/Pr RI RI RI RI RI RI RI RI/Pr RI/Pr RI RI RI/Pr RI/Pr RI RI RI RI/Pr RI Diabetes mellitus Y N N N N N N N N N N N N N N N N N N N N N N N Hypertension Y Y N Y Y N N N N N N N Y Y N N N Y N N Y N N N Cr (mg/dl) 2.3 2.3 4.2 3.9 3.4 1.9 2 2.1 2.2 1.3 3.1 3.3 3 3.8 8 2 2.4 3.8 1.4 1.7 1.9 3.7 1.8 2 CCr (ml/min) 48 55 16 25 U 34 28 33 20 68 U 27 U U U U U 32 78 35 U U U U 24-h protein (g)b 6.7 10.5 0.351 6 1.5 4.8 0.048 0.912 0.54 0.02 “1ϩ” 0.4 0.7 2 3 “0” 0.18 2.7 4.5 “0” 0.3 0.654 2.5 “0” Salb (g/dl) 2.8 3.2 U 4 3.7 3.6 U 3.9 4.2 U 4 U U U NL 4.4 4.6 NL 3.2 3.8 4.6 U NL U Schol (mg/dl) 355 412 U 274 207 322 U 246 175 U 223 U U U U 352 198 216 232 167 260 U U U Edema Y N N N N N U N U N N N U U N N N N N N N U N U Bipolar disorder Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Years on lithium 15 14 12 20 “Many” 14 11 5 10 9 15 2 10 20 18 20 9 13 20 8 12 10 20 25 Apparent NDI Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y U N Y N N Y Y Y Follow-up d/c lithium Y Y Y Y Y Y Y Y Y Y Y Y Y N Y Y U Y Y U Y U Y U length follow-up 6 yr 5 mo 13 mo 14 mo 14 mo 3 mo 3 mo 1 yr 3 yr 8 mo 22 mo 1 yr 7 yr 7 mo HD U U 5 yr 2 yr U 2.5 yr U 11 yr U Cr 2.2 2.2 5/HD 6.4/H 6.1/HD 1.4 1.9 1.3 2.4 0.8 4.8 3.7 6/HD HD HD U U HD 1.7 U 1.9 U 7.6/HD U Light microscopy CTIN c/w Liϩ toxicity Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y

FSGS lesions Y Y N Y Y Y N N N N N N N Y Y Y N Y Y N Y N N 1441 Y Nephropathy Progressive A Nephrotoxicity: Lithium other diagnosis N N AIN N N N N N U ThinBM N N N ?FSGS N N N N N N N N N N no. of glomeruli 22 26 5 22 25 36 20 10 6 13 4 16 38 86 87 13 47 18 12 20 53 10 13 Ͼ100 no. of segmentally 24 0 3 2 2 0 0 0 0 0 00 13710220 4 0 0 20% sclerotic glomeruli no. of globally sclerotic 9 17 3 13 5 31 6 9 4 3 1 10 27 58 65 8 31 13 9 14 20 9 6 20% glomeruli TA/IFc Mod Sev Sev Sev Sev Sev Mod Sev Sev Mod Sev Mod Mod Sev Sev Sev Sev Mod Sev Mod Sev Mod Mild Mod tubular cysts Y Y N Y DIL DIL Y Y DIL Y Y Y DIL Y DIL DIL Y Y Y Y Y DIL DIL Y vascular diseasec Mod Mod Mild Mod Mild Mod Mild Mod U Mild Mild Mod Mod Mod Mod Mild Mild Mod Mild Mild Mod Mild NL Mod Immunofluorescence Neg Neg Neg Neg Neg Neg Neg Neg Neg Neg Neg Neg Neg Neg Neg Neg Neg Neg Neg Neg Neg Neg Neg Neg EM, deposits U N U U N U U U N N U N N N N N N U U N N N N N EM, FPF% U 90 U U 60 U U U 10 10 U 80 0 U 60 30 60 U U 15 20 0 30 10

a W, white; AA, African-American; H, Hispanic; RI, renal insufficiency; Pr, proteinuria; Y, yes; N, no; Cr, creatinine; CCr , creatinine clearance; U, unknown; Salb , serum albumin; Schol , serum cholesterol; NL, normal; NDI, nephrogenic diabetes insipidus; d/c, discontinue; HD, hemodialysis; CTIN, chronic tubulointerstitial nephropathy; c/w, consistent with; FSGS, focal segmental glomerulosclerosis; AIN, acute interstitial nephritis; ThinBM, thin basement membrane disease; TA/IF, tubular atrophy and interstitial fibrosis; Mod, moderate; Sev, severe; DIL, dilated (tubular profiles); Neg, negative; EM, electron microscopy; FPF, foot process fusion. b Quotation marks around values for dipstick proteinuria indicate that 24-h protein quantification was not performed. c Pathologic scoring is graded as mild (involving on average 0 to 25% of cortex or vessel walls), moderate (26 to 50%), or severe (Ͼ50%). 1442 Journal of the American Society of Nephrology J Am Soc Nephrol 11: 1439–1448, 2000

Table 2. Patient demographics Table 4. Renal biopsy findingsa

Total no. of patients 24 Light microscopy Patient age, yr (mean) 42.5 (range, 26 to 57) CTIN 24 of 24 100% Male/female 12/12 FSGS 12 of 24 50% Race TA/IF Caucasian 21 of 24 87.5% mild 1 of 24 4.2% African-American 2 of 24 8.3% moderate 9 of 24 37.5% Hispanic 1 of 24 4.2% severe 14 of 24 58.3% Diabetes mellitus 1 of 24 4.2% Tubular cysts 15 of 24 62.5% Hypertension (at time of 8 of 24 33.3% Tubular dilatation without cysts 8 of 24 33.3% biopsy) Immunofluorescence, negative 24 of 24 100% Bipolar disorder 24 of 24 100% EM, glomerular deposits 0 of 15 0% Mean duration of lithium 13.6 (range, 2 to 25) EM, Ͼ50% FPF 5 of 14 35.7% therapy (yr) Additional diagnosesb 3 of 24 12.5%

a Abbreviations as in Table 1. b AIN, ? primary FSGS, thin basement membrane disease. Table 3. Renal presentationa

Renal insufficiency 24 of 24 100% ments to the tubular basement membrane by cytoplasmic ex- Mean SCr initial (mg/dl) 2.8 (range, 1.3 to 8.0) tensions from neighboring cells (Figure 3C). There was focal Proteinuria 10 of 24 41.7% apoptosis of individual cyst lining cells (Figure 3D). Multilay- Ͼ ( 1 g/24 h) ering of tubular epithelia was observed rarely. The tubular 24-h urine protein (g) basement membrane displayed foci of irregular lamellation 0 to 0.5 10 of 24 41.7% with hyaline insudation and focal detachment of tubular epi- 0.5 to 1.0 4 of 24 16.7% thelial cells with neomembrane formation (Figure 3C). 1.0 to 3.0 4 of 24 16.7% Each biopsy was carefully examined for the presence of Ͼ3.0 6 of 24 25% b glomerular disease. Global glomerulosclerosis was observed in Nephrotic syndrome 3 of 24 12.5% all biopsies and was common in zones of chronic tubulointer- Clinical evidence of NDI 20 of 23 87% stitial damage. The mean percentage of glomeruli that were a Abbreviations as in Table 1. globally sclerotic was 57.5% (range, 20 to 90%; median b Defined as three of the following four criteria: proteinuria Ͼ3 63.9%). In 12 of the 24 biopsies (50%), lesions of FSGS were Ͻ Ͼ g/d, Salb 3.3 g/dl, Schol 200 mg/dl, and peripheral edema. identified (Table 4, Figure 4, A and B). Within these 12 biopsies, lesions of FSGS were seen on average in 11.5% of glomeruli (range, 5.6 to 20%; median 10.1%). These lesions with TP. In some cysts, the lectin markers gave discontinuous were typically characterized by loss of patency of the glomer- staining, with skip areas (Figure 2D). Thus, approximately half ular capillary lumen, increased matrix material, hyaline insu- of the tubular cysts stained positively for EMA and negatively dation, occasional endocapillary foam cells, adhesions to Bow- for AH, and the other half stained positively for both EMA and man’s capsule, and, in some cases, mild hypertrophy and AH. These staining patterns correspond to distal tubule and hyperplasia of overlying visceral epithelial cells. Glomerulo- collecting duct origin, respectively. In contrast, tubular atrophy megaly was identified in the majority of cases (77.3%). No affected all three nephron segments. biopsy displayed evidence of proliferative glomerulonephritis. Ultrastructural evaluation of cystic and noncystic distal tu- Immunofluorescence was negative except for focal nonspecific bular epithelia revealed widespread alterations in tubular epi- glomerular and vascular positivity for IgM and C3 in areas of thelial fine structure, predominantly affecting distal and col- glomerulosclerosis and arteriosclerosis. In the 15 cases in lecting tubules. Major alterations included loss or which glomerular tissue was available for ultrastructural anal- simplification of apical microvilli, apical blebbing, dilation of ysis, no electron-dense deposits were seen. The degree of foot the endoplasmic reticulum producing numerous electron lucent process fusion ranged from 0 to 90% (mean 34%), with Ͼ50% vacuoles, increased number of phagolysosomes, decreased or- effacement in five of 14 cases (35.7%) (Figure 4C). Podocyte ganellar content, and a paucity of cytoplasmic organelles in the microvillous transformation was commonly seen in cases with perinuclear region, producing perinuclear clearing (Figure 3, A Ͼ50% foot process effacement. and B). Mitochondria often appeared swollen with rare prox- Three biopsies were accorded a second diagnosis in addition imal tubular giant mitochondria. Nuclei were typically dis- to lithium-associated CTIN. Patient 10 had a history of micro- placed apically. There was disorganization of the basolateral hematuria and had diffuse thinning of the glomerular basement infoldings of plasma membrane, at times extending up to the membrane (mean 175 nm), consistent with thin basement apical tight junctions. The basal width of epithelial cells was membrane disease. Patient 3 had clinical evidence of pyelone- often irregular, with focal undermining of epithelial attach- phritis, and a renal biopsy displayed an acute polymorphonu- J Am Soc Nephrol 11: 1439–1448, 2000 Lithium Nephrotoxicity: A Progressive Nephropathy 1443

Table 5. Immunohistochemical and lectin staining of tubular cystsa

Total No. EMA Case of Cysts Positive AH Positive TP Positive

7221 0 10 1 1 0 0 11 3 3 3 1 12 3 3 3 2 14 10 10 10 0 20 3 3 3 1 21 3 3 2 0 24 80 80 32 0 Total 105 105 (100%) 54 (51.4%) 4 (3.8%)

a EMA, epithelial membrane antigen; AH, Arachis hypogaea; TP, Tetragonolobus purpureas. clear and granulomatous interstitial nephritis superimposed on a Discussion chronic tubulointerstitial nephropathy. An intriguing case was Many clinical and pathologic studies have examined the scope patient 14, who presented with subnephrotic proteinuria and a of lithium-associated CTIN. The first major analysis that included serum creatinine of 3.8 mg/dl. An open renal biopsy displayed pathologic correlation was published in 1977 and described 14 severe tubular atrophy and interstitial fibrosis, many tubular cysts, patients who presented with either acute intoxication or NDI (3). and multiple glomeruli with lesions of FSGS. The findings were Although the indication for biopsy was not chronic renal insuffi- believed to be typical of lithium-associated CTIN. The patient had ciency, the main pathologic findings were tubular atrophy and a past history of attempted suicide when lithium was temporarily interstitial fibrosis (3). Subsequently, 69 patients receiving lithium withdrawn and, as a result, discontinuation of lithium was not therapy for Ͼ2 yr were examined; 18 patients (26.1%) developed possible. Two months after the biopsy, the patient became fully a renal concentrating defect and eight patients (11.6%)—all of nephrotic and 5 mo later he progressed to ESRD and was placed whom were treated with lithium for at least 4.5 yr (mean 9.3 on hemodialysis. The patient subsequently received a renal trans- yr)—developed an increase in serum creatinine of Ͼ0.3 mg/dl plant, and his posttransplant course was complicated by the de- during the course of treatment (13). Another study followed 57 velopment of severe nephrotic syndrome at 2 mo posttransplant patients receiving lithium and found that five individuals (8.8%), due to biopsy-documented recurrence of FSGS. Throughout this all of whom had been treated for Ͼ8 yr, went on to develop a time, the patient continued to receive lithium, suggesting the “moderately decreased” GFR during the course of therapy (4). possibility of recurrent lithium-induced FSGS, although primary This study provides limited data on the 24 subjects who under- FSGS could not be excluded. went renal biopsy, but the major findings were focal interstitial Clinical follow-up was available in 19 of the 24 cases (Table fibrosis and global glomerulosclerosis (4). A retrospective study 6, Figure 5). The mean follow-up time was 30.6 mo, with a of 142 patients receiving lithium for Ͼ15 yr found a reduction of maximum of 11 yr. In all but one patient, lithium was discon- the GFR in 21% of patients (15). A separate study examined 46 tinued either before or immediately after the renal biopsy. One renal biopsies performed on patients receiving lithium for a mean patient committed suicide after discontinuation of lithium. De- of 8 yr and found significantly greater tubular atrophy, interstitial spite lithium withdrawal, eight of the 19 patients progressed to fibrosis, and glomerulosclerosis than in biopsies from age-matched ESRD (42.1%). By Cox regression analysis, the only signifi- control subjects not treated with lithium (primarily patients with cant predictor of progression to ESRD was initial serum cre- proteinuria and “normal” renal biopsies) (5). A prospective study of atinine at the time of biopsy (P ϭ 0.060). Among the 10 65 patients receiving lithium showed a significant decline in creati- patients with an initial serum creatinine Ͻ2.5 mg/dl, only one nine clearance as early as 1 yr after initiation of therapy in the 26 male progressed to ESRD and that occurred 11 yr later. Furthermore, subjects only (16). A review of lithium nephrotoxicity included data three of the 10 patients experienced a decline in their serum from 14 separate studies and found that 85% of lithium-treated creatinine of Ն0.5 mg/dl. In contrast, seven of the nine patients patients have a normal GFR and that the remaining 15% have only with an initial serum creatinine of Ͼ2.5 mg/dl went on to mild reductions in renal function (7). require hemodialysis, and an additional patient had a rise in A major point of contention has been whether lithium ther- serum creatinine from 3.1 to 4.8 mg/dl over a 22-mo period of apy can lead to marked reductions in renal function or ESRD. follow-up (Figure 5). By Kaplan–Meier survival estimates, a Aurell et al. examined renal biopsies in six patients with serum creatinine of Ͼ2.5 mg/dl at the time of biopsy was a established renal insufficiency who had been treated with lith- significant predictor of progression to ESRD (P ϭ 0.008) ium for 4 to 13 yr and found tubular atrophy, interstitial (Figure 6), while the presence of FSGS on renal biopsy was not fibrosis, sclerotic glomeruli, and tubular microcysts (6). Unlike predictive of progression (P ϭ 0.14) (data not shown). many of the previous reports, some of the patients in this study 1444 Journal of the American Society of Nephrology J Am Soc Nephrol 11: 1439–1448, 2000

Our retrospective, biopsy-based study is not a cross-sectional analysis and therefore cannot address the prevalence of lithium nephrotoxicity among lithium-treated patients. The biopsy prevalence of lithium nephrotoxicity is low, representing 0.37% of all native renal biopsy diagnoses in the 14-yr period of 1986–1999. All 24 patients presented with renal insufficiency and many also had significant proteinuria. Indications for renal biopsy were not standardized. In general, referring nephrologists stated that they follow multiple patients on lithium and biopsy those individuals manifesting an “atypical course” or “significant” degrees of renal insufficiency or proteinuria. Two open biopsies (patients 14 and 24) were obtained at the time of nephrectomy for oncocytoma and papillary renal cell carcinoma, respectively. Lithium-associated CTIN is characterized by tubular atrophy and interstitial fibrosis, typically out of proportion to the extent of glomerular or vascular disease. In our 24 cases, a detailed clinical history, light microscopy, immunofluorescence, and electron microscopy were available for review. In each case, the possibility of a separate etiology for the renal abnormali- ties, such as glomerulonephritis or hypertensive arterionephro- sclerosis, was excluded. Three patients received a second diagnosis, including thin basement membrane disease, interstitial nephritis, and possible primary focal segmental glomerulosclerosis (recurrent in transplant). The association of renal neoplasms in two of the cases, both in the setting of chronic lithium nephrotoxicity with multifocal cysts, raises the ques- tion of a possible increased risk of renal neoplasia in this population, as may occur in other cystic diseases. However, an increased risk of renal neoplasms has not been reported in larger cross-sectional studies. Although CTIN represents a somewhat nonspecific pattern of disease, the presence of tubular cysts is highly characteristic of lithium toxicity, having been reported in up to 40% of cases (13). New Zealand white rabbits treated with lithium develop a pattern of CTIN with tubular cysts that is virtually identical to the human disease, with progressive renal insufficiency (17). Male Wistar rats treated with lithium develop nephrogenic diabetes insipidus and a distal tubulopathy marked by tubular dilatation (18). We were able to find tubular cysts in 62.5% of biopsies and lesser degrees of tubular dilatation in an additional 33.3% of biopsies. Previous studies have not always documented this important and relatively specific finding (4). Of Figure 1. (A) Severe lithium-associated chronic tubulointerstitial ne- note, some studies have attempted to deny the existence of phropathy (CTIN) with diffuse interstitial fibrosis and tubular atrophy lithium-associated CTIN by describing similar pathologic find- and relative sparing of glomeruli (patient 5). Trichrome stain, ϫ40. (B) Another example of severe lithium-associated CTIN with the ings in patients with affective disorders who have not received additional finding of focal tubular cysts arising in a background of lithium (19,20). Yet the one pathologic finding that was seen severe interstitial fibrosis and tubular atrophy (patient 24). Periodic only in the lithium-treated group is tubular cysts (20), under- acid-Schiff (PAS) stain, ϫ40. (C) High-power view of tubular cysts scoring the specificity of this morphologic finding. lined by simple cuboidal epithelium (labeled C). Adjacent tubules The potential glomerular toxicity of lithium therapy has been show tubular dilatation (labeled d) (patient 24). PAS stain, ϫ100. underappreciated. Although lesions of FSGS have been described in only three patients on lithium therapy (11), we had marked degrees of renal insufficiency. Of note, one patient identified lesions of FSGS in 50% of the 24 biopsies on lithium presented with a GFR of 13 cc/min, which improved to 25 therapy. There was a strong correlation between the finding of cc/min 3 yr after discontinuation of therapy (6). A single case FSGS and the presence of proteinuria Ͼ1.0 g/d (P ϭ 0.0014, of lithium-associated CTIN leading to ESRD has been reported Fisher exact test). A potential explanation is that the patients in the literature (8). with higher initial serum creatinine levels had lost a significant J Am Soc Nephrol 11: 1439–1448, 2000 Lithium Nephrotoxicity: A Progressive Nephropathy 1445

Figure 2. Panel of lectin and immunohistochemical staining of tubular cysts. Two cysts from patient 21 (A, hematoxylin and eosin) stained negatively for Tetragonolobus purpureas (TP) (B), positively in one of two cysts with Arachis hypogaea (AH) (C), and positively in both cysts for epithelial membrane antigen (EMA) (D). Staining for EMA is discontinuous with segmental skip areas along the cyst circumference (D). A representative field from patient 24 contains multiple cysts that stain negatively for TP (E), focally positive for AH (arrow in F), and diffusely positive for EMA (G). Magnification: ϫ100 in A through D; ϫ40 in E through G. 1446 Journal of the American Society of Nephrology J Am Soc Nephrol 11: 1439–1448, 2000

Figure 3. (A) A cystic tubule shows complete loss of apical brush border, focal dilatation of endoplasmic reticulum, and reduced number of organelles, especially in the perinuclear region. (B) High-power view of a collecting tubular epithelial cell shows apical displacement of the nucleus with complete loss of apical differentiation and prominent perinuclear clearing. (C) A distal nephron segment displays diffuse dilatation of endoplasmic reticulum and overlapping cellular interdigitations along the basal and apical aspects. The tubular basement membrane is irregularly lamellated. (D) A distal nephron segment with a single cell demonstrating nuclear condensation and probable early apoptosis (arrow). The adjacent cells have prominent apical nuclei and complete loss of surface differentiation. Magnification: ϫ2500 in A and D; ϫ8000 in B; and ϫ3000 in C.

proportion of their renal parenchyma, causing hyperfiltration in This would not be altogether surprising in light of the well- remnant nephrons and development of secondary FSGS. The established occurrence of lithium-associated minimal change high prevalence of glomerulomegaly (77.3%), a relatively con- disease (9–10). Although the high degree of proteinuria in our stant feature of the secondary FSGS mediated by structural- study reflects in part the indications for biopsy, this potential functional adaptations to loss of functioning nephrons, favors manifestation of lithium nephrotoxicity has not been recog- this possibility (21). Nonetheless, a few lines of evidence argue nized previously. A report of 44 biopsies in patients with against this hypothesis. First, there was no correlation between lithium-associated renal disease in which the mean proteinuria the degree of renal insufficiency and either the 24-h urine was 250 mg/d described no case of proteinuria exceeding 1000 protein or the presence of FSGS. Second, the presence of FSGS mg/d (20). The presence of significant foot process fusion in did not correlate with the severity of tubular atrophy and some cases and the recurrence of FSGS in the single transplant interstitial fibrosis. Finally, there was a high incidence (35.7%) patient in whom lithium was not withdrawn lend further sup- of Ͼ50% foot process effacement, an uncommon finding in port to the hypothesis of potential direct glomerular toxicity. secondary FSGS mediated by hyperfiltration injury (22). An important new finding in our cohort of patients is that Therefore, an alternative hypothesis must be considered. Al- lithium-associated CTIN can lead to profound renal insufficiency though some cases of FSGS in the setting of lithium-associated and even ESRD despite discontinuation of therapy. Seven of the CTIN may occur secondary to the tubulointerstitial changes, 18 patients in whom follow-up was available went on to develop others may reflect a direct glomerular (i.e., podocyte) toxicity. ESRD, and one patient presented with ESRD and a serum creat- J Am Soc Nephrol 11: 1439–1448, 2000 Lithium Nephrotoxicity: A Progressive Nephropathy 1447

Table 6. Clinical outcome dataa

Follow-up available 19 of 24 79.2% Mean follow-up time (months) 30.6 (range, 3 to 132) Lithium discontinued 20 of 21 95.2% End-stage renal disease 8 of 19 42.1%

SCr Initial SCr Initial Ͻ2.5 mg/dl Ͼ2.5 mg/dl (n ϭ 10) (n ϭ 9) Ͼ SCr declines 0.4 mg/dl 30 Ͻ SCr rises or declines 0.5 mg/dl 61 Ͼ SCr rises 0.4 mg/dl (or ESRD) 18 a Abbreviations as in Table 1.

Figure 5. Course of renal disease in patients with lithium nephrotoxicity. Changes in serum creatinine from the time of biopsy are displayed. Solid lines indicate patients with evidence of FSGS, and dashed lines represent patients without FSGS.

serum creatinine of Ͼ2.5 mg/dl had progression of disease as opposed to one of 10 patients with a serum creatinine Ͻ2.5 mg/dl (P ϭ 0.008, Kaplan–Meier survival estimates). One should not infer from these data that a serum creatinine Ͻ2.5 mg/dl is necessarily safe because some of the patients had discontinued Figure 4. (A) A typical example of focal segmental glomerulosclero- lithium at some time before biopsy, when they had even lower sis (FSGS) shows a discrete lesion of sclerosis with focal hyalinosis serum creatinine levels. The development of ESRD in patient 12 and adhesion to Bowman’s capsule (patient 21). PAS stain, ϫ250. (B) who received lithium for only 2 yr and discontinued lithium 10 yr A glomerulus from the native kidney biopsy of patient 14, who before biopsy exemplifies this point. developed recurrent FSGS in the allograft, contains a typical lesion of Immunohistochemical and lectin staining revealed tubular segmental sclerosis with mild podocyte hypertrophy. PAS stain, cysts of predominantly distal tubular (EMA ϩ/AH Ϫ) and ϫ250. (C) Ultrastructural evaluation of a nonsclerotic glomerulus collecting duct (EMA ϩ/AH ϩ) origin. Although these find- from patient 12 reveals extensive foot process fusion and focal mi- ings had been predicted based on the predominantly distal ϫ crovillous transformation of the podocytes. Magnification: 2500. nephron toxicity of lithium, ours is the first documentation of the anatomic sites of cystogenesis. The rare cysts expressing lectin TP, in addition to EMA and AH, may represent proximal inine of 8.0 mg/dl. The only reliable predictor of progression to propagation of cysts from more distal portions of the nephron. ESRD was the serum creatinine at the time of biopsy (P ϭ 0.060, In conclusion, our biopsy-based study in a selected popula- Cox regression analysis). Eight of nine patients with an initial tion of lithium-treated patients is not inconsistent with the 1448 Journal of the American Society of Nephrology J Am Soc Nephrol 11: 1439–1448, 2000

4. Jorgensen F, Larsen S, Spanager B, Clausen E, Tango M, Brinch E, Brun C: Kidney function and quantitative histologic changes in patients on long-term lithium therapy. Acta Psychiatr Scand 70: 455–462, 1984 5. Hetmar O, Brun C, Clemmesen L, Ladefoged J, Larsen S, Rafaelson OJ: Lithium: Long-term effects on the kidney. II. Structural changes. J Psychiatr Res 21: 279–288, 1987 6. Aurell M, Svalander C, Wallin L, Alling C: Renal function and biopsy findings in patients on long-term lithium treatment. Kid- ney Int 20: 663–670, 1981 7. Boton R, Gaviria M, Batlle DC: Prevalence, pathogenesis, and treatment of renal dysfunction associated with chronic lithium therapy. Am J Kidney Dis 10: 329–345, 1987 8. Hestbech J, Aurell M: Lithium-induced uraemia. Lancet 1: 212– 213, 1979 Figure 6. Kaplan–Meier survival curve in patients dichotomized for 9. Alexander F, Martin J: Nephrotic syndrome associated with initial serum creatinine more than or less than 2.5 mg/dl. Serum lithium therapy. Clin Nephrol 15: 267–271, 1981 creatinine Ͼ2.5 mg/dl at the time of biopsy is a significant predictor 10. Tamm VKK, Green J, Schwieger J, Cohen AH: Nephrotic syn- of progression to end-stage renal disease (P ϭ 0.008). drome and renal insufficiency associated with lithium therapy. Am J Kidney Dis 27: 715–720, 1996 11. Santella RN, Rimmer JM, MacPherson BR: Focal segmental consensus opinion that only a minority of patients receiving glomerulosclerosis in patients receiving lithium carbonate. Am J lithium will develop mild renal insufficiency due to lithium- Med 84: 851–854, 1988 associated CTIN. Furthermore, the importance of lithium for 12. Nadasdy T, Laslik Z, Blick KE, Johnson DL, Silva FG: Tubular patients with affective disorders is underscored by the fact that atrophy in the end-stage kidney: A lectin and immunohistochem- one patient (patient 2) who discontinued lithium subsequently ical technique. Hum Pathol 25: 157–163, 1994 13. Hansen HE, Hestbech J, Sorensen JL, Norgaard K, Heilskov J, committed suicide. However, our data indicate for the first time Amdisen A: Chronic interstitial nephropathy in patients on long- that a minority of patients receiving lithium will develop sig- term lithium therapy. Q J Med 48: 577–591, 1979 nificant renal insufficiency with characteristic renal biopsy 14. Burrows GD, Davies B, Kincaid-Smith P: Unique tubular lesions findings of CTIN and distal/collecting tubular cysts, and that after lithium. Lancet 1: 1310, 1978 these patients are at risk for progression to ESRD despite 15. Bendz H, Aurell M, Balldin J, Mathe AA, Sjodin I: Kidney prompt withdrawal of lithium. The only predictor of progres- damage in long-term lithium patients: A cross-sectional study of sion to ESRD is the initial serum creatinine. These findings patients with 15 years or more on lithium. Nephrol Dial Trans- emphasize the need for physician vigilance and regular peri- plant 9: 1250–1254, 1994 odic measurements of renal function for early detection of 16. Jorkasky DK, Amsterdam JD, Oler J, Braden G, Alvis R, Geheb lithium-induced CTIN. Serum creatinine levels in the range of M, Cox M: Lithium-induced renal disease: A prospective study. 2.0 mg/dl, generally considered to represent only “mild” de- Clin Nephrol 30: 293–302, 1988 grees of renal insufficiency among non-nephrologists, may 17. Walker RG, Escott M, Birchall I, Dowling JP, Kincaid-Smith P: represent a level of inexorably progressive renal injury. Fi- Chronic progressive renal lesions induced by lithium. Kidney Int nally, our new finding of a high prevalence of FSGS and 29: 875–881, 1986 nephrotic proteinuria in this biopsy population, independent of 18. Kling MA, Lox JG, Johnston SM, Tolkoff-Rubin NE, Rubin RH, the level of renal insufficiency, suggests a potential direct Colvin RB: Effects of long-term lithium administration on renal glomerular toxicity. structure and function in rats. Lab Invest 50: 526–535, 1984 19. Kincaid-Smith P, Burrows GD, Davies BM, Holtwill B, Walther References M, Walker RG: Renal biopsy findings in lithium and prelithium 1. Timmer RT, Sands JM: Lithium intoxication. J Am Soc Nephrol patients. Lancet ii: 700–701, 1979 10: 666–674, 1999 20. Walker RG, Bennett WM, Davies BM, Kincaid-Smith P: Struc- 2. Marples D, Christensen S, Christensen EI, Ottosen PD, Nielsen tural and functional effects of long-term lithium therapy. Kidney S: Lithium-induced downregulation of aquaporin-2 water chan- Int 21[Suppl 11]: S13–S19, 1982 nel expression in rat kidney medulla. J Clin Invest 95: 1838– 21. Rennke HG, Klein PS: Pathogenesis and significance of non- 1845, 1995 primary focal and segmental glomerulosclerosis. Am J Kidney 3. Hestbech J, Hansen HE, Amdisen A, Olsen S: Chronic renal Dis 13: 443–456, 1989 lesions following long-term treatment with lithium. Kidney Int 22. D’Agati V: The many masks of focal segmental glomeruloscle- 12: 205–213, 1977 rosis. Kidney Int 46: 1223–1241, 1994

Access to UpToDate on-line is available for additional clinical information at http://www.jasn.org/