Nephrogenic diabetes insipidus
The most common renal side effect of lithium is of concentrating urine despite normal or elevated concentrations of the antidiuretic hormone vasopressin (Table
1). The concentrating defect leads to decreased urine osmolality and increased urine volume (polyuria). A urinary concentrating defect may occur without overt polyuria, but this is usually clinically insignificant because near-normal urinary volume rarely leads to the performance of urine osmolality tests.
Table 1
Controversies over lithium and the kidney (selected references)
Lithium causes anatomic kidney damage | | Vestergaard et al. ( 1979) |
Low-dose lithium prevents kidney damage | | |
Duration of lithium treatment predicts Li polyuria | | |
Duration of lithium predicts reduced GFR | | |
The incidence of nephrogenic diabetes insipidus (NDI) among lithium-treated patients varies greatly in different studies, with a prevalence range of 20 to 87 % (Baylis and Heath
1978; Vestergaard et al.
1979; Vestergaard and Amdisen
1981; Schou and Vestergaard
1988; Okusa and Crystal
1994; McKnight et al.
2012; Markowitz et al.
2000; Kallner and Petterson
1995; Bucht and Wahlin
1980; Boton et al.
1987). For example, in a comprehensive meta-analysis of studies comprising 1172 lithium-treated patients, Boton et al. (
1987) reported a reduction in urinary concentrating ability in more than 54 % of patients while only 19 % had overt polyuria.
Many studies found that major factors affecting the incidence and severity of urinary concentrating defects among lithium-treated patients are as follows: duration of treatment (longer duration increases the risk of NDI), blood lithium level, and frequency of acute lithium intoxication
s (Vestergaard et al.
1979; Vestergaard and Amdisen
1981; Schou and Vestergaard
1988; Markowitz et al.
2000; Bucht and Wahlin
1980; Boton et al.
1987; Walker
1993; Turan et al.
2002; Timmer and Sands
1999; Bendz et al.
2001). On the other hand, Lepkifker et al. (
2004) observed that the duration of lithium treatment and plasma concentrations were not associated with increased risk of urinary concentrating defect while episodes of lithium intoxication were more predictive and positively correlated. A urinary concentrating defect may occur as early as 2 to 4 months after the commencement of lithium (Boton et al.
1987; Smigan et al.
1984), but it becomes more evident after chronic treatment (Vestergaard and Amdisen
1981; Markowitz et al.
2000; Bendz et al.
2001; Phillips et al.
2008). NDI may develop even after cessation of lithium therapy (Paw et al.
2007). Additionally, several studies demonstrated that cessation of long-term lithium therapy does not always restore the urinary concentrating capacity of the kidney (Markowitz et al.
2000; Bendz et al.
2001; Khairallah et al.
2007) while others reported that it may alleviate the concentrating defect (Bucht and Wahlin
1980). These variations may be related to the stage of tubulointerstitial damage at which lithium was stopped. An early stage at which only
functional tubulointerstitial damage has occurred may be fully reversible. On the other hand, a late stage at which irreversible
morphological tubulointerstitial changes (fibrosis) have occurred will not be resolved by lithium cessation. Importantly, patients with polyuria that do not consume sufficient amounts of fluids are at a high risk of becoming volume-depleted, further increasing the risk of lithium toxicity (Vestergaard et al.
1979; Vestergaard and Amdisen
1981; Smigan et al.
1984). Furthermore, patients who were concurrently treated with lithium and neuroleptic drugs had significantly higher rates of urinary concentrating defects than patients who were treated only with lithium. Consistently, antipsychotics (only)-treated patients had a higher incidence of urinary concentrating defects than matched healthy subjects (Bucht and Wahlin
1980). These observations suggest that antipsychotic drugs may contribute to the development of urinary concentrating defects.
Acute lithium treatment reduces the antidiuretic effect of vasopressin (Singer et al.
1972). Similarly, chronic lithium treatment was shown to reduce the antidiuretic effect of vasopressin by several possible mechanisms. The most established pharmacological treatment for lithium-induced NDI is the potassium-sparing diuretic amiloride (Boton et al.
1987; Timmer and Sands
1999; Feuerstein et al.
1981; Bedford et al.
2008). However, amiloride is likely to be effective only when there is a mild to moderate urinary concentrating defect that is potentially reversible. Among the suggested mechanisms by which amiloride alleviates lithium-induced urinary concentrating defect are as follows:
i) blocking of epithelial sodium channel ENaC, resulting in reduced absorption of lithium into collecting duct cells (Walker et al.
1982), and
ii) restoring to normal the expression levels of AQP2 and AQP3 (Bedford et al.
2008).
Chronic kidney disease
According to the American National Kidney Foundation (
2002), chronic kidney disease (CKD) is defined as either kidney damage or GFR <60 ml/min/1.73 m
2 for ≥3 months. Kidney damage is defined as pathologic abnormalities or markers of damage, including abnormalities in blood or urine tests or imaging tests. The staging of CKD according to GFR (ml/min/1.73 m
2) is as follows: Stage 1—kidney damage with normal or increased GFR (GFR ≥90); Stage 2—kidney damage with a mildly decreased GFR (GFR = 60–89); Stage 3—moderate reduction in GFR (GFR = 30–59); Stage 4—severe reduction in GFR (GFR = 15–29); and Stage 5—kidney failure (GFR <15 or dialysis) (American National Kidney Foundation
2002). End-stage renal disease (ESRD) describes a situation in which patients need dialysis or transplantation, irrespective of their level of kidney function (American National Kidney Foundation
2002).
Early studies questioned the association between lithium and chronic impairment in glomerular function (Vestergaard et al.
1979; Vestergaard and Amdisen
1981; Boton et al.
1987; Walker
1993; Walker et al.
1982; Bendz et al.
1983,
1994; Coskunol et al.
1997; Paul et al.
2010). In 1979, Vestergaard et al. (
1979) examined renal function among 237 patients who had been treated with lithium for 0.5–17 years (average 5 years). They determined GFR by examining 24-h creatinine clearance and serum creatinine. It was found that lithium treatment reduced GFR only slightly and that the risk of developing ESRD was low (Bucht and Wahlin
1980). Two years later, these authors published a follow-up report on 184 patients of the same population in which they re-examined renal function (Vestergaard and Amdisen
1981). For 37 patients, lithium had been discontinued while the other 147 were still on lithium. None of the patients in either group had a reduction in GFR (Vestergaard and Amdisen
1981). Impairments in urinary concentrating ability had however progressed in the lithium-treated group. Subsequently, Walker et al. (
1982) reported that the results of renal biopsy samples analyzed for interstitial fibrosis showed no difference between lithium-treated patients and those who did not receive lithium. However, abnormal serum creatinine levels were significantly higher in lithium-treated patients as compared to those who did not receive lithium, suggesting an impaired GFR (Walker et al.
1982). A meta-analysis study by Boton et al. revealed that 15 % of 1172 patients on chronic lithium therapy displayed only mild reduction in GFR (Timmer and Sands
1999). Bendz et al. (
1983) examined glomerular function in outpatients who received short-term and long-term lithium treatment. They found that 3 % had abnormal glomerular function while 51 % had impaired tubular function. Eleven years after the previous study (Bendz et al.
1983), they reported that among 142 patients who received lithium for more than 15 years, a reduction in GFR was observed in 21 % of the patients (Bendz et al.
1994). The results also revealed that co-administration of lithium with other psychotropic drugs or somatic medications increased the risk of altered renal function. Coskunol et al. (
1997) compared 107 lithium-treated patients to 29 non-lithium-treated patients matched for age and sex. They found that plasma creatinine levels and creatinine clearance did not significantly differ between the two groups. In addition, they found that there was no significant association between creatinine clearance and the duration (or dosage) of lithium treatment (Coskunol et al.
1997). A meta-analysis of 23 studies conducted by Paul and co-workers concluded that: “any lithium-associated increase in serum creatinine is quantitatively small and of questionable clinical significance;” this is despite a significant increase in serum creatinine levels in several studies that were analyzed (Paul et al.
2010). A comprehensive meta-analysis of 385 studies by McKnight et al. (
2012) found a non-significant (
p = 0.15) reduction in GFR and a low risk for developing ESRD among lithium-treated patients.
Despite the uncertainty reported in the studies cited above, other studies have suggested that long-term lithium therapy increases the risk of CKD (Markowitz et al.
2000; Turan et al.
2002; Bendz et al.
2001,
2010; Aiff et al.
2014a; Kripalani et al.
2009; McCann et al.
2008; Presne et al.
2003). One of the studies that demonstrated deleterious and mostly irreversible renal damage after chronic lithium therapy was that by Markowitz et al. (
2000), which included 24 patients (average lithium treatment duration 13.6 years) with renal insufficiency. Renal biopsy results revealed a chronic tubulointerstitial nephropathy in all patients, with associated cortical and medullary tubular cysts and dilation in 62.5 and 33.3 % of patients, respectively (Markowitz et al.
2000). Nine of 19 patients progressed to ESRD despite lithium withdrawal (eight of nine had an initial serum creatinine above 2.5 mg/dL and one of ten had an initial serum creatinine below 2.5 mg/dL). On the other hand, three patients with initial creatinine levels below 2.1 mg/dL had subsequent improvement in renal function (Markowitz et al.
2000). The authors concluded that serum creatinine levels higher than 2.5 mg/dL are the only significant predictor of progressing to ESRD.
The Markowitz et al. (
2000) study was retrospective, biopsy-based from a selected renal biopsy database. Of 6514 biopsies that made up the database, only 24 (0.37 %) were those of lithium-treated patients. Moreover, it is not clear whether lithium therapy was indeed the sole reason for kidney disease and referral to renal biopsy. One of the patients who discontinued lithium subsequently committed suicide, underscoring the complexity of discontinuing lithium even in the presence of kidney disease. A study by Presne et al. (
2003) also demonstrated the correlation between long-term lithium therapy and CKD. It included 54 patients for whom chronic lithium treatment was apparently the only cause of renal disease (during the study period, six other patients were excluded because they had evidence of other causes for renal disease). Another 20 patients on chronic lithium therapy were referred for systematic renal biopsy. Among those 74 patients (mean age at onset of lithium therapy 42.9 years, mean duration of lithium therapy 19.8 years), 12 (16%) reached ESRD at a mean age of 65 years (Presne et al.
2003). Renal function was determined by measuring serum creatinine and creatinine clearance. Lithium-induced chronic nephropathy seemed to be a slowly progressing disease. The average latency between initiation of lithium and the progression to ESRD was estimated at 20 years (Presne et al.
2003). The severity of interstitial fibrosis on renal biopsy was associated with treatment duration and cumulative dose of lithium.
A survey of lithium-induced ESRD in French dialysis centers was conducted. Among 10,726 patients surveyed, 24 lithium-treated patients (0.22 %) had ESRD (Presne et al.
2003). The authors found that the duration of lithium treatment is the most significant factor that negatively affects renal function. A cohort study of kidney damage in long-term lithium patients demonstrated a significant correlation between treatment duration and the frequency of decreased GFR and urinary concentrating ability (Bendz et al.
2001). Eighty-six patients were on lithium (mean age 58, mean treatment duration 16 years) and 42 patients (mean age 57, mean treatment duration 10 years) were off lithium at follow-up time. For lithium-treated patients, the frequency of reduced urinary concentrating ability and increased serum creatinine significantly increased during mean follow-up periods of 8 and 10 years, respectively (Bendz et al.
2001). The authors concluded that among patients on long-term lithium therapy, the prevalence of significant nephropathy increases with time and that reduced GFR is less common than a reduction in urinary concentrating ability (Bendz et al.
2001). Similarly, a study by McCann et al. (
2008) examined 38 patients who had been treated with lithium for a mean of 6.9 years (short-term) and 21 patients who were on lithium for a mean of 14.2 years (long-term). They found that long-term treatment patients had a significantly higher serum creatinine and urea levels (McCann et al.
2008). However, this study had several limitations including the relatively small sample size and being a retrospective, cross-sectional study. These and other limitations led the authors to emphasize that although “regression analysis does show that longer duration of lithium use is associated with higher creatinine level,…, this is not necessarily associated with a clinically relevant abnormalities in renal function” (McCann et al.
2008).
Bendz et al. (
2010) assessed the prevalence of ESRD that required renal replacement therapy (RRT) in a lithium-treated population in Sweden. They found that 2202 subjects were on RRT among the surveyed general population (2,684,157), a prevalence of 0.8/1000. On the other hand, among a population of 3369 lithium-treated patients, 18 were on RRT (5.3/1000), representing a sixfold increase in RRT prevalence (Bendz et al.
2010). For the latter 18 patients on RRT, lithium treatment was apparently the only reason for ESRD (lithium-treated patients on RTT were excluded if they had other reasons that may have caused ESRD). The average age at the start of RRT among those 18 patients was 62 years (range 46–74), and the average duration of lithium treatment before initiation of RTT was 23 years (range 12–35). Moreover, of the 3369 lithium-treated patients, 41 had plasma creatinine levels above 150 μmol/L (not including RRT patients), a prevalence of 12.2/1000. The authors concluded that the duration of lithium therapy was the only risk factor for developing CKD and ESRD. Long-term treatment was defined as a duration of ≥15 years on lithium. It was found that 16 (89 %) of the lithium-treated RTT patients had received lithium for more than 15 years. Similarly, lithium-treated patients who had plasma creatinine levels above 150 μmol/L were older than the general population of lithium-treated patients (average age 68 versus 56 years, respectively). The results of this retrospective study (Bendz et al.
2010) support a positive correlation between long-term lithium therapy and development of CKD (Aiff et al.
2014b).
Reduced GFR may sometimes occur at early stages after initiation of lithium therapy. For example, a prominent reduction in GFR was found among 53 patients who received lithium for only 4 months (Smigan et al.
1984). Lepkifker et al. (
2004) studied 114 patients who had been taking lithium for 4 to 30 (mean 16.75) years. The control group included 94 untreated non-psychiatric subjects matched for sex and age (mean duration of follow-up 11.9 years). Ninety lithium-treated patients (79%) showed no increase in creatinine levels during a follow-up of up to 30 years and were similar to the control group. The other 24 patients had a progressive increase in creatinine levels during the follow-up period and were classified as having renal insufficiency (Lepkifker et al.
2004). The development of renal insufficiency was associated with episodes of lithium intoxication, other diseases, and/or drugs that affect renal function but not with the duration of lithium treatment in a majority of patients (Lepkifker et al.
2004).
The above reviewed evidence seems contradictory and difficult to summarize. Aiff et al. (
2014a) suggested a potential solution to the contradictions: They found using their large Swedish registry that no patient treated since 1980 has developed ESRD in their sample and concluded that low-dose treatment apparently in use since then in Sweden does not cause ESRD. This is supported by the results of Aprhamian et al. (
2014) who found no changes in kidney function after 2 years of low-dose lithium in elderly patients. However, Close et al. (
2014) did not find this “time of treatment” effect in the UK in their large registry study. This “historical” approach does not explain the very early studies that found no effects of lithium in kidney biopsy. A general recommendation to use low-dose lithium, while consistent with the medical principle of
primum non nocere and medical caution, does not help the clinician with difficult patients who respond fully only to high-dose maintenance. High-dose lithium has been reported to have increased efficacy in both prophylaxis and bipolar depression. Clearly, individual patient history of response and severity of episodes must be critical in the decision-making (Belmaker et al.
2012).
In summary, the evidence attests to a small but definite risk of lithium-induced reduction in GFR and progression to CKD. Some studies indicate that the risk for CKD and ESRD is low while others suggest that long-term lithium treatment increases the risk for chronic nephropathy to a clinically relevant degree.