Lithium has been a subject reported on in this Journal for a very long time. ^{Reference Sainsbury1} Perhaps unsurprisingly, in more recent decades most of the reports about the clinical use of salts of this monovalent cation have concerned its use as a treatment for affective disorders and despite its status as a generic medicine, new information about the benefits of lithium to health continues to be published. As consideration of these data has matured, early doubts about lithium's utility as a treatment ^{Reference Moncrieff2} have been replaced by a solid and mature evidence base upon which we can confidently base clinical practice. A good example of this is the meta-analytic evidence which shows that the therapeutic use of lithium may prevent suicide. ^{Reference Young and Hammond3} Intriguing, though still not entirely conclusive, new evidence is published on this topic in the current issue. ^{Reference Kapusta, Mossaheb, Etzersdorfer, Hlavin, Thau and Willeit4} For the treatment of bipolar disorder, lithium has good evidence for antimanic actions, although these benefits must be balanced against the likelihood of rebound mania upon stoppage. ^{Reference Goodwin5} The value of lithium, both as monotherapy and combined with valproate, for relapse prevention in bipolar disorder was also shown in the recent BALANCE study. ^{Reference Geddes, Goodwin, Rednell, Azorin, Cipriani and Ostacher6} In contrast, lithium monotherapy performed disappointingly as an acute treatment for bipolar depression in a recent trial. ^{Reference Young, McElroy, Bauer, Philips, Chang and Olausson7} However, this lack of an antidepressant effect may possibly have been due to the relatively short-term (8 weeks) duration of this study. However, just when it seemed that knowledge of lithium and its benefits for neuropsychiatric disorders had reached an uncontroversial consensus, a twist in this tale has appeared. Studies, initially from the realm of affective disorders, produced new and, perhaps somewhat surprising, evidence that lithium may confer benefits in terms of preventing dementia.

Lithium, bipolar disorder and dementia

Nunes et al ^{Reference Nunes, Forlenza and Gattaz8} compared the prevalence of Alzheimer's disease in 66 elderly euthymic patients with bipolar disorder who were on chronic lithium therapy with 48 similar patients who had not received lithium therapy recently. These case–control data suggested that lithium reduced the prevalence of Alzheimer's disease in patients with bipolar disorder to that found in the general elderly population. Nunes et al ^{Reference Nunes, Forlenza and Gattaz8} discounted this effect of lithium as a result of a reduction in the number of affective episodes, as in their study this variable was equivalent between the lithium and non-lithium groups. They suggested rather that this putative effect of lithium was due to ‘a result of its intrinsic biological effects in the brain’ and speculated that the effect of lithium to inhibit the transcription of the glycogen synthetase kinase (GSK)-3 gene might be pivotal. Indeed, lithium had previously been shown to produce a relative inhibition of GSK-3-induced tau phosphorylation. ^{Reference Lovestone, Davis, Webster, Kaech, Brion and Matus9} Subsequent studies confirmed that patients with affective disorders on long-term lithium had a reduced risk for Alzheimer's disease, albeit admitting that, because of the non-randomised nature of the data, these findings might be spurious. ^{Reference Kessing, Sondergard, Forman and Anderson10,Reference Kessing, Forman and Anderson11} Convincing proof of lithium's ‘antidementia’ properties would therefore require data from appropriately designed, prospective, randomised trials. A small, open-label study examined the effects of administering lithium carbonate to patients with Alzheimer's disease for up to 1 year. ^{Reference Macdonald, Briggs, Poppe, Higgins, Velayudhan and Lovestone12} Disappointingly, no cognitive benefits were found in the patients who completed the study. In a subsequent placebo-controlled, single-blind study, lithium sufate was used to treat patients with mild Alzheimer's disease for 10 weeks. Lithium treatment was not found to have significant benefits on either cognitive performance or on cerebrospinal fluid (CSF) concentrations of disease-related biomarkers. ^{Reference Hamel, Ewers, Burger, Annas, Mortberg and Bogstedt13}

Thus, it might appear that lithium has joined the many compounds which, despite great promise at the earlier stages of evaluation, failed to show efficacy in clinical trials. However, two points are of note before we write lithium off on the basis of these data. First, these are only two small and preliminary trials. Larger trials, perhaps with different dosages of lithium for a different duration of time, may produce more positive results. Second, it may be inappropriate to test drugs in patients with fully manifest, albeit mild, Alzheimer's disease. Studies of patients at high risk may be more fruitful. Such an approach was taken by Forlenza et al, ^{Reference Forlenza, Diniz, Radanovic, Santos, Talib and Gattaz14} members of the research group which had produced one of the original papers about the protective effects of lithium. ^{Reference Nunes, Forlenza and Gattaz8} The main objective of this study was to assess the effect of long-term lithium treatment on the progression of cognitive deficits in patients with amnestic mild cognitive impairment (aMCI), a state which has a high rate of progression to frank Alzheimer's disease. In this study, the effect of lithium on CSF concentrations of putative Alzheimer's disease biomarkers was also measured. Forty-five individuals with aMCI were randomised to receive lithium (0.25–0.5 mmol/l) or placebo in a double-blind trial of 12 months duration. A significant decrease in CSF concentrations of phospho-tau and a better performance on cognitive measures was found in the lithium-treated group. The number of patients who progressed from aMCI to Alzheimer's disease was higher in the placebo group (7 out of 20) than in the lithium treated group (4 out of 21), although this difference was not statistically significant. Clearly, this trial is encouraging and the effects of lithium on cognition and Alzheimer's disease biomarkers are very suggestive of likely benefit. The numbers progressing to Alzheimer's disease were lower, although albeit not significantly so, presumably because of the limited power of the study. This trial adds to the increasing evidence that lithium may have beneficial effects on the brain ^{Reference Macritchie, Lloyd, Bastin, Vasudev, Gallagher and Eyre15} and begs to be replicated in further randomised trials in patients with aMCI.

Where do we go from here?

Further trials will be required and these should incorporate the features of this study, which may have been related to its success. These include a long duration of treatment and a study of patients with aMCI rather than full-blown Alzheimer's disease. The dose of lithium used is also of note. We tend to assume that the lithium levels required for therapeutic effect are those derived from studies of mania. ^{Reference Severus, Lipkovich, Licht, Young, Greil and Ketter16} In the present study, markedly lower levels were induced (0.25–0.5 mmol/l) for a much longer time period. The notion that lower levels of lithium for a longer period of time may have benefits for human health has received some recent support. In a study from Japan, Ohgami et al, ^{Reference Ohgami, Terao, Shiotsuki, Ishii and Iwata17} demonstrated that lithium levels in drinking water were significantly and negatively associated with suicide rates. These findings suggest that even very low levels of lithium in drinking water may play a role in reducing suicide risk within the general population. Such findings, when considered with the evidence from Forlanza et al ^{Reference Forlenza, Diniz, Radanovic, Santos, Talib and Gattaz14} raise the possibility that environmental lithium may also prevent dementia and clearly suggest that epidemiological studies examining this should be carried out.

What about treatment?

The pharmaceutical industry is clearly very much focused on developing treatments for dementia, although results to date have been equivocal and no disease-modifying agents are either licensed or can be currently recommended for clinical use. ^{Reference O'Brien and Burns18} If such treatments were to be given to large numbers of at-risk individuals for prolonged periods of time, the commercial rewards to those owning the patents for such treatments are likely to be very considerable. Lithium of course is under no patent and will not attract industry funds for further development as a treatment except perhaps as a comparator to commercial compounds or possibly in combination with another agent. The onus is therefore on governmental and charitable funding agencies. Such trials will not be cheap, but were they to prove positive, the possible benefits in health to our ever-ageing population would be beyond any such price.