This observational study provides a snapshot of current prescription attitudes regarding lithium treatment in the elderly at our center.
Dosing and maintenance of serum concentrations
Maintenance doses in this elderly cohort (median 450 mg of lithium carbonate daily) were much lower compared to the usual dosage of 900–1500 mg of lithium carbonate per day used in younger adults (Baldessarini and Tarazi
2006). Such doses are consistent with the results from the McGill Geriatric Lithium-Induced Diabetes Insipidus Clinical Study (Rej et al.
2014b) reporting that the lithium dose required to achieve a given serum concentration decreases threefold from middle to old age. With regard to the therapeutic range, serum concentrations are not well established in the elderly. In our lithium clinic, the lower end of the range for elderly patients is reduced to 0.40 mmol/l compared to the 0.60 mmol/l generally indicated for younger adults (Bauer and Gitlin
2016), principally to prevent episodes of lithium toxicity associated with reduced renal function, drug interactions, and risk of dehydration, all of which are typical of old age.
Moreover, there is evidence that the ratio between serum and brain concentration may decrease with age (Moore et al.
2002).
Comedication
The high proportion of patients from this cohort taking regularly psychiatric and non-psychiatric medications warrants some comments. For example, it is perhaps surprising that 3/4 of our patients were on long-term benzodiazepines. It is possible that the specialists prioritized the need to control anxiety and/or sleep disorders over the widely known detrimental side effects of benzodiazepines on cognition (Larson et al.
1987). Another point of note is that medications that are theoretically contraindicated in lithium patients (ACE inhibitors, sartans, and thiazides) were taken regularly by more than half of the patients in this cohort. The likely explanation here is that hypertension must be treated even in lithium patients, provided that the doses are reduced adequately. This notion is corroborated by the fact that patients in this cohort treated with ACE inhibitors or sartans were taking significantly lower lithium doses. Moreover, with regard to cases with reduced eGFR, the decision to treat patients with ACE inhibitors or sartans was often taken by the nephrologist, based on the evidence that these drugs may slow the progression of CKD (Ruggenenti et al.
2012). We have not included data on other nephrotoxic drugs, namely NSAIDs, because they were not taken regularly but only when needed.
Diagnoses
A subgroup of patients in this cohort who began taking lithium when elderly recalls the case series of 10 patients described by Hagop Akiskal’s group of “elderly patients with late-onset mood and related behavioral symptomatology and cognitive decline without past history of clear-cut bipolar disorder”. The authors reported that “symptoms were often refractory to or aggravated by antidepressants…, whereas mood stabilizers and/or atypical antipsychotics were beneficial, promoting behavioral improvement in all treated patients and marked cognitive recovery in five”. Only one case had been treated with lithium (Ng et al.
2008). We did not measure cognition in the present study, but the above-mentioned subgroup had a history of behavioral dysregulation, mood lability, and cognitive symptoms that improved after withdrawing long-term antidepressant treatment and starting lithium. According to referring specialists, their choice of prescribing lithium was also based on its neuroprotective properties (Bauer et al.
2003). Interestingly, a longer duration of lithium treatment has been reported to be related to higher white matter integrity among elderly patients with bipolar disorder (Gildengers et al.
2015). In another study using voxel-based morphometry, Zung et al. (
2016) found an increased left hippocampal volume in the lithium-treated group compared with the nonlithium-treated group, and decreased left hippocampal volume in the nonlithium group relative to controls.
Thyroid function
One-third of patients from this cohort were taking thyroid hormone replacement therapy, including two patients who were prescribed levothyroxine for the first time during follow-up. It has long been known that lithium is associated with hypothyroidism, and that middle-aged women are at higher risk (Johnston and Eagles
1999). In the aforementioned cross-sectional study of thyroid function by Kraszewska et al. (
2015), of 45 women receiving long-term lithium (mean age, 63 years), seven (16%) were receiving levothyroxine replacement therapy. In three patients, this drug had been introduced within the 1st year of lithium therapy, and in the remaining four patients after 8, 11, 12, and 33 years.
Hyperthyroidism requiring antithyroid medication was less frequent in patients from this cohort (4%), and manifested for the first time during follow-up in two cases.
Therefore, we conclude that thyroid function continues to require close monitoring throughout lithium treatment.
Renal function
The problem of lithium-associated renal dysfunction has recently been addressed by several research studies and reviews (Rej et al.
2012,
2015). With regard to old patients, a series of studies from the University of Toronto have provided several pertinent results. In a 4-year retrospective cohort study of 42 patients, lithium levels did not correlate with change in eGFR, suggesting that levels up to 0.8 mmol/l are safe in geriatric patients without pre-existing chronic renal failure (Rej et al.
2013a). In a 5-year retrospective cohort study of 27 geriatric patients with eGFR lower than 60 ml/min/1.73 m
2, changes in eGFR in patients who continued lithium for at least 2 years and those who discontinued lithium did not differ significantly. However, clinically important decreases in eGFR occurred in the majority of continuers but in none of the discontinuers (Rej et al.
2013b). In a 4-year retrospective cohort study of 82 patients, geriatric psychiatry patients were found at higher risk for clinically important decreases in eGFR than 200 psychotropic-naïve similarly aged controls. Multivariate analyses of potential risk factors for renal dysfunction (including age, hypertension, diabetes, diuretics, and duration of lithium treatment), suggested that lithium is an important factor when eGFR is lower than 60 ml/min/1.73 m
2 (Rej et al.
2014c). In a population-based cross-sectional study of 2480 lithium users aged ≥70 years, lithium use for >2 years was one of the factors independently associated with CKD, together with hypertension, diabetes mellitus, ischemic heart disease, nephrogenic diabetes insipidus, acute kidney injury, and use of loop diuretics, hydrochlorothiazide, or atypical antipsychotics (Rej et al.
2014b).
We have previously reported that duration of lithium treatment is to be added to advancing age as a risk factor for reduced glomerular filtration rate (Bocchetta et al.
2013,
2015). In a retrospective regression analysis of the last available eGFR regarding 953 (596 women, 357 men) patients of any age, eGFR was found lower in women (by 3.47 ml/min/1.73 m
2), in older patients (0.73 ml/min/1.73 m
2 per year of age), and in patients with longer lithium treatment (0.73 ml/min/1.73 m
2 for each year) (Bocchetta et al.
2015). In the present study, the effects of sex and age were not significant because of the small sample size and the limited age range.
The decline attributable to lithium was 0.85 ml/min/1.73 m
2 per year of prior exposure. During follow-up, the decline was faster, corresponding to an annual decline of 2.3 ml/min/1.73 m
2. It must be noted that 39% of patients had already an eGFR lower than 60 ml/min/1.73 m
2 at baseline, and the initial mean eGFR in patients followed up for 6 years was 66.4 ml/min/1.73 m
2. There are data suggesting that, once CKD is established, eGFR further declines irrespective of lithium withdrawal (Bendz et al.
2010; Bocchetta et al.
2015). A similar pattern of decline is also observed in the general population: for example, a longitudinal analysis performed on 4074 subjects from the Sardinia study cohort (Pani et al.
2014) revealed that eGFR declined by 1.87 ml/min/1.73 m
2 per year in individuals with a baseline eGFR of <60 ml/min/1.73 m
2 compared to a decline of 0.80 ml/min/1.73 m
2 per year in individuals with a baseline eGFR of ≥60 ml/min/1.73 m
2.
In a 2-year randomized, placebo-controlled trial followed by single-blind extension, lithium treatment was not associated with renal dysfunction (Aprahamian et al.
2014). The trial was carried out in the context of a study of long-term lithium treatment for amnestic mild cognitive impairment in the elderly (Forlenza et al.
2011). It must, however, be noted that the target of lithium concentrations was low (0.25–0.50 mmol/l), and subjects with CKD might have been excluded from the trial, because enrollment was reliant on the approval of the general practitioner.
A recent population-based study concluded that, after adjustment for several confounders (including comorbidities, co-prescriptions, and episodes of lithium toxicity), the decline attributable to lithium was not significant (Clos et al.
2015). However, this study did not include patients older than 64 years, and patients with an eGFR <60 ml/min/1.73 m
2 represented only 1% of their cohort.
Limitations of the study
The noninterventional nature of this study implies several limitations. Apart from ordinary clinical procedures, no specific instruments were used to evaluate response to treatment or side effects. Cognition was not measured, and neuroimaging was not performed systematically. With regard to the role of lithium exposure in the decline of renal function, the absence of a control group has to be taken into account. Moreover, several potential risk factors for CKD (such as hypertension, diabetes, co-medications) that can be detected in population-based samples were not adjusted for because of the small size of this clinical sample.