More robust evidence for the efficacy of lithium in the long-term treatment of bipolar disorder: should lithium (again) be recommended as the single preferred first-line treatment?

Among the many options for the long-term treatment of bipolar disorder, lithium is
the oldest drug. After its initial use by the Danish neurologist Lange in the nineteenth
century and its subsequent discovery by the Australian psychiatrist Cade in 1949,
the efficacy of lithium was finally established by Schou and his Danish colleagues
in the 1960s which led to its registration in Europe and the US as well as many other
countries in the 1970s (Bech 2006]). Since then, several other drugs have become available as alternatives, of which
the anticonvulsants carbamazepine, valproate and lamotrigine and several atypical
antipsychotics have also been registered for the long-term treatment of bipolar disorder.
Subsequently, all registered drugs also found their place in guidelines, supported
by data from randomised placebo-controlled clinical trials and systematic reviews
and meta-analyses, including a meta-analysis on lithium versus placebo which found
that lithium did prevent any mood episodes and manic episodes, but was not significantly
better than placebo in the prevention of depressive episodes (Geddes et al. 2004]).

Overall, these data led to recommendations in most guidelines for more than one drug
as first-line treatment for the long-term treatment of bipolar disorder. For instance,
the latest US APA guideline from 2002 (!) recommends lithium or valproate and as possible
alternatives lamotrigine, carbamazepine, oxcarbazepine and ongoing antipsychotic treatment
(American Psychiatric Association 2002]); the UK NICE guideline from 2006 lithium, olanzapine or valproate (National Institute
for Health and Care Excellence (NICE) 2006]); the updated WFSBD guideline from 2012 ‘based on strongest level of evidence for efficacy and a good risk–benefit ratio’ aripiprazole, lamotrigine, lithium and quetiapine (Grunze et al. 2013]); and finally the updated Canadian CANMAT and ISBD guideline from 2013 lithium, valproate,
olanzapine, quetiapine, lamotrigine, risperidone long-acting injection (LAI) and ziprasidone
(Yatham et al. 2013]).

In 2011, the so far largest long-term study on the efficacy of lithium was published
(Weisler et al. 2011]). In this double-blind trial, bipolar I patients who had responded to and had tolerated
quetiapine were randomised to continue with quetiapine or to switch to lithium or
placebo. The results showed that both quetiapine and lithium did prevent any mood
episodes, manic episodes and depressive episodes. Quetiapine was also more effective
than lithium (in the prevention of any mood episodes and depressive episodes), but
this result should be interpreted with caution as the study design was enriched for
quetiapine (see above) but not for lithium. Only the WFSBP and the CANMAT/ISBD guidelines
did include the Weisler et al. 2011 trial, while the WFSBP guideline also concluded
that lithium was the only drug with the strongest level of evidence (level A for any
mood episodes, manic episodes and depressive episodes) in non-enriched samples.

Two recent meta-analyses further looked at the evidence of lithium. In this journal,
Severus et al. (2014]) published an update and extension of the meta-analysis by Geddes et al. from 2004.
With seven placebo-controlled trials, i.e. two more than in 2004 including the Weisler
et al. 2014 trial, they now found that lithium not only prevents any mood episodes
and manic episodes but also prevents depressive episodes. This finding was further
supported by a network meta-analysis including 33 randomised controlled trials examining
17 treatments (both monotherapies and combination therapies) by Miura et al. (2014]). Of the monotherapies, only lithium and quetiapine were found to prevent any mood
episodes, manic episodes and depressive episodes when compared to placebo, with the
notion that the evidence for lithium is based almost completely on trials with a non-enriched
design and for quetiapine only on trials with enriched designs. Both meta-analyses
also looked at tolerance of lithium (more dropouts due to adverse events with lithium
than with placebo) and at overall satisfaction with treatment (more completers with
lithium than with placebo).

The conclusions of both meta-analyses are rather similar. Severus et al. state that
‘with no other drug available having such ample and consistent evidence for its efficacy,
lithium remains the most valuable treatment option in this indication’, and Miura et al. conclude that ‘lithium seems to be the most reasonable candidate for a first–line option for the long–term treatment of bipolar disorder’. This raises the question whether lithium should indeed be recommended as the single
preferred first-line treatment. Indeed, the recent 2014 update of the NICE guideline
(with their own meta-analyses) recommends to ‘offer lithium as a first–line, long–term pharmacological treatment for bipolar disorder’?, while regarding the other alternatives, the guideline continues as follows: ‘if lithium is ineffective, consider adding valproate or if lithium is poorly tolerated,
or is not suitable (for example, because the person does not agree to routine blood
monitoring), consider valproate or olanzapine instead or, if it has been effective
during an episode of mania or bipolar depression, quetiapine’. Thus, despite these considerations, the National Institute for Health and Care
Excellence (NICE) (2014]) guideline recommends lithium as the first option.

One of the major limitations of all meta-analyses is that while they looked at efficacy,
tolerability and overall satisfaction, their scope was not more than 2 years. However,
when choosing a drug for long-term if not life-long treatment, one should look at
efficacy and safety not only during the first years but also thereafter. The use of
lithium over more than 10 years is associated with the risk of kidney dysfunction
and the use of atypical antipsychotics with metabolic syndrome and an increased mortality
risk due to cardiovascular problems. Although it has not yet been established how
these late adverse effects have an impact on the long-term safety of lithium and the
atypical antipsychotics as its major alternatives, the assumption is that also over
10 or more years the safety of lithium is at least in balance with that of most atypical
antipsychotics. However, it would be interesting to see studies in patients with bipolar
disorder looking at mortality during treatment with lithium and its alternatives over
more than 10 years such as has already been done with clozapine versus other antipsychotics
in patients with schizophrenia (Tiihonen et al. 2009]).

In the meantime, I concur with the conclusions of both meta-analyses supporting the
recommendation in the NICE guideline that lithium, given its evidence base from trials
with a non-enriched design and its relative safety, should be the single first-line
treatment. Whether or not this conclusion will be supported by most experts in the
field of lithium will become clear in 2016 when the recently started ISBD task force
on the use of lithium will present its report.