Lithium

In 1812 Benjamin Rush observed that “Many mad people, who have attempted to destroy themselves by cutting their throats . . . have been cured by the profuse haemorrhages.” Blood letting soon became his first remedy for mania.¹ With the advent of a more scientific approach to medicine, this treatment, based on centuries of tradition and glowing clinical testimonials, fell into well deserved disrepute. Should lithium follow it?

Almost 140 years after Rush’s observations John Cade noted that the toxicity of urine injected into guinea pigs was attenuated by lithium. After finding that lithium had “no discernable ill effects” when he took it himself, Cade successfully treated 10 manic patients with the drug.² Thus, in 1949, the modern era of lithium therapy began. Almost simultaneously, however, the ill advised use of lithium chloride as a salt substitute for patients on low sodium diets produced reports of neurotoxicity and death.³ From this beginning the battle lines were drawn. Was this simple element a safe and effective cure for various ills or an ineffective but toxic nostrum?

In 1968 Blackwell and Shepherd suggested that lithium prophylaxis for bipolar disorder was “another therapeutic myth”⁴—the mid-20th century equivalent of blood letting. Subsequently several placebo controlled trials proved successful and the critics were silenced—but only temporarily. Recently, after reconsidering the “evidence,” Moncrieff concluded that lithium may be ineffective not only for prophylaxis but also for acute mania and for augmentation in treatment resistant depression.⁵ Her assertions have not gone unchallenged⁷ and, in general, lithium remains a highly valued treatment modality.⁸

The early double blind, placebo controlled studies of lithium for acute mania were less than ideal in design, but, despite the drawbacks (which probably minimised the differences between lithium and placebo), lithium consistently outperformed placebo. Not until recently, however, was the antimanic efficacy of lithium confirmed in a large (n=179), multicentre, parallel design, placebo controlled study.⁹ The impact of rescue medication was minimised by limiting it to chloral hydrate or lorazepam during the first 10 days of the three week study and never within eight hours of behavioural assessment. Based on an intention to treat analysis, 50% or greater improvement occurred in 49% of patients on lithium (n=36), 48% on divalproex (n=69), and 25% on placebo (n=74). Prior lithium treatment had been ineffective in 42% of those randomised to lithium, so the the deck was stacked against a favourable response to lithium, yet it emerged clearly superior to placebo. Nevertheless, in clinical practice few would argue that lithium alone is an adequate treatment for other than the milder cases of mania.

While lithium maybe an imperfect long term treatment for bipolar disorder, it is difficult to embrace Moncrieff’s conclusion that it is ineffective. The abrupt withdrawal of lithium in placebo controlled discontinuation studies may have exaggerated drug-placebo differences because of withdrawal induced mania. Subsequent comparisons of affective morbidity with rapid (1-14 day) versus gradual (15-31 day) lithium discontinuation found more rapid and higher recurrence rates in the former group.¹⁰ To use these observations to argue for the ineffectiveness of lithium prophylaxis is specious, however, because even with gradual discontinuation recurrence rates were high.

When Goodwin and Jamison reviewed 10 placebo controlled maintenance studies, they found a relapse rate difference of 47% in favour of lithium.¹¹ A meta-analysis of placebo controlled studies found a 55% difference in relapse rate favouring lithium which reached a statistical significance of p<10^-29.¹² There is growing, although not incontrovertible, evidence that lithium prophylaxis reduces mortality in patients with bipolar disease (particularly from suicide).¹³

Poorer responses to lithium have been associated with mixed or dysphoric mania, rapid cycling, many previous episodes, impaired functioning between episodes, and a depression-mania-euthymia course. While anticonvulsants such as carbamazepine and valproate (and more recently gabapentin and lamotrigine) show promise in these areas, properly designed studies have not compared their efficacy with lithium. At present alternatives to lithium are welcome options when lithium is ineffective or not tolerated, but it is questionable whether they should displace lithium as treatments of first choice.

The addition of lithium to an antidepressant to overcome treatment resistant depression has become an established intervention with about 50% effectiveness. Controlled studies continue to support this approach.¹⁴,¹⁵ It is difficult to share Moncrieff’s scepticism, especially when studies she reported as negative actually had positive outcomes.¹⁵,¹⁶

All drugs have side effects, and lithium is no exception. In overdose it is toxic, and deaths and permanent neurological and renal damage have occurred. Even at therapeutic levels, lithium commonly causes polyuria and impaired renal concentrating ability, and there is growing evidence that a minority of patients experience a gradual reduction in glomerular filtration rate which is probably caused by lithium.¹⁷ Thus, periodic measurement of serum creatinine and, when indicated, 24 hour urine volume, protein, and creatinine clearance have become an integral part of long term lithium management.

Lithium is neither a paragon of therapeutic perfection nor a highly toxic placebo. Rather, it has established clinical utility for acute mania, for prophylaxis of bipolar (and probably unipolar) disorder, and for augmentation in treatment resistant depression. While we should continue to seek more effective and safer treatments, until they arrive the words of Ambroise Paré remain pertinent: “Better a tried remedy than a new fangled one.”¹⁸