Lithium toxicity

Lithium toxicity in the CICM exams fits into"Poisoning and drug intoxication", a massive topic that does most of the heavy lifting for Section 2.1.14 in the second edition of the CICM Syllabus for the Second Part Examination. Considerting that this ancient dirty drug remains incredibly popular, it is surprising that this has never appeared in the past papers until Question 10 from the first paper of 2024. Prior to this it had been a footnote in data interpretation SAQs like Question 3.1 from the second paper of 2013 and Question 29.1 from the first paper of 2021. 

Oruch et al (2014) is probably the best single reference for the toxicology of lithium, as the CICM exam question was heavily invested in the different manifestations of toxicity, and most papers do not list them in quite the same level of detail as Oruch does. However most normal people would go directly to the Toxicology Handbook by Murray et al, where one can find everything one needs in a clear succinct style on pages 279-284 of the 3rd ed. And for those of us who have no access to a current ED trainee, the lithium chapter by Lavonas & Brent (2017) from Critical Care Toxicology will be sufficient. Wherever it is not otherwise stated, the following summary was derived mostly from these two sources.

Why this monovalent cation?

Lithium is one of the few drugs that has remained the gold standard for the management of its original indication since it was first introduced in 1949. We really have nothing better to treat bipolar disorder, and it is so effective that experts call for "organizational changes ...to systematically and long-term change the use of lithium on a large-scale population-level" (Kessing, 2024). The evidence base to support the use of this drug is much larger than the literature to support any other drug. Without overextending into an unfamiliar specialty, the author can point to papers such as Nestsiarovich et al (2022), systematic reviews of copious trials comparing lithium with newer agents (valproate, lamotrigine, quetiapine, etc etc) have yielded consistently pro-lithium findings, and it remains a first line agent for bipolar disorder and mania. Though this is not directly the objective of this chapter, a discussion of the clinical effects might interest some readers, and they are redirected tho the excellent works by Lenox & Hahn (2000), Quiroz et al (2004) and Pasquali et al (2010). In summary, the mechanisms of action include:

• By pretending to be magnesium (their ionic radius is similar), lithium inhibits several enzymes, in particular ones involved in recycling and synthesis of inositol for the inositol-1,4,5-triphosphate (IP₃) signalling pathways, but also numerous others:
  • adenylate cyclase
  • phosphoinositide
  • arachidonic acid
  • GSK-3, a serine–threonine kinase 
  • FBPase, BPNase, and PGM
  • Neurotrophin signalling pathway enzymes
• Though papers will often extensively elaborate on these pathways as playing a critical role in this thing or that, nobody can commit to which of these produces the desirable neuroprotective and behavioural effects.

If monovalent cation is the key characteristic, then why not others? That might seem silly, but the reader is directed to see their nearby Periodic Table of Elements, where monovalent cations are arrayed along the left hand edge. One will observe that they are all in use.  Hydrogen is lighter than lithium but impractical for home use, sodium and potassium are the next heaviest but are somewhat overinvolved in biological processes already, which means the next available monovalent cation is, unfortunately, rubidium.  Yes, reader, before you ask, rubidium was in fact used as a mood stabiliser, and appears to have had some effect  - for example, Paschalis et al (1978) reported some clinical experience with it - but it never became as popular as lithium, perhaps because of the cost. Though it is strictly speaking more abundant than zinc or copper in the Earth's crust, the act of extracting it is more laborious because it reacts with everything and has charming tendency to ignite spontaneously in contact with air and water.

Epidemiology of lithium toxicity

Shuy et al (2024), a Singaporean writing about data collected mainly not from Australian sources, noted a decline in the use of this drug to treat bipolar disorder from 27% to 17% in the US, but a steady hum of activity in Europe (~35%) and a slight increase in Asia (from 25% to 26%).  Considering these values to be representative, and taking seriously the findings by the Australian Bureau of Statistics that 2.9% of Australians (~568,000 people) have bipolar disorder, and the finding that in 2022-2023,  0.3% of all Australians were dispensed at least one lithium prescription, one's simple maths suggest that somewhere between 96,560 and 198,800 people are taking an average dose of 600mg bd, which amounts to 116,000-239,000 kg of lithium daily, or 42,340-87,235 tonnes per every calendar year. These data cannot possibly be true, because then this would represent about a third of the total annual lithium production in 2016, or approximately 0.01% of the total lithium reserves in the lithosphere. In short, though the CICM syllabus calls for epidemiology ("For each of the above L1 conditions expected knowledge will include", etc), it is sometimes dangerous to ask for this, because some weirdos will take it literally.

From the above, strange statements notwithstanding, one can extract the one specific finding, namely that from the ED presentation data and our common shared experience of practicing in this field, we all conclude that lithium use is highly prevalent, whereas lithium toxicity is not; and so it must be a reasonably safe drug. Even though virtually all patients on lithium will experience some side effects at one point or another, only a small fraction will ever appear in the emergency department, and of these only a smaller fraction yet again will require the attention of an intensivist. Ott et al (2016), writing in Europe (Sweden) where lithium prescribing is very popular, were able to scrape together only 1340 patients over seventeen years, of whom only 77 experience episodes of toxicity, and only 26 required ICU admission, making it a once-every-eight-months sort of ICU presentation in the 200,000-strong population of Norrbotten. Extrapolating these data from that fjord-rich environment to the vast arid plains of Australia, one might expect one to turn up every month or so to Westmead ICU, which services a population of 1,850,000. 

Aetiology of lithium toxicity

To include "Aetiology" in the expected knowledge for toxicology topics might seem disingenuous ("its because they took all the tablets, that's the causative factor"), but in reality most lithium toxicity does have an antecent cause entirely inrelated to intentional self-harm. The drug has a narrow therapeutic index and it is possible to diligently overdose yourself while taking the correct dose you have been prescribed. Of the associated factors, Ott et al (2016) observed the following:

• Infection was the precipitant in about 10% of cases
• Many patients developed toxicity after being prescribed an interacting drug, eg. NSAID or thiazide (these interfere with lithium elimination)
• Many developed a decline in renal function unrelated to lithium, and then went on to accumulate lithium.

Pathophysiology of lithium toxicity

And at this stage, even though pharmacokinetics and pharmacodynamics are meant to remain in the First Part Exam, a digression into the pharmacology of lithium needs to occur, most because it is key to the differences between the toxicity that results from acute ingestion of a large amount of lithium, and the chronic slow accumulation of lithium that occurs over the course of weeks and months, leading to a very different clinical syndrome. 

Clinical course of acute and chronic lithium toxicity

Acute lithium toxicity  is a short term affair, often self-inflicted, and fortunately often consequence-free. It is well absorbed orally and following a large acute overdose one may manifest some concerning symptoms within a couple of hours (Okusa & Crystal, 1994). Chronic toxicity is usually much slower in onset (days, weeks) and is equally slow to resolve, with some symptoms potentially never going away at all. 

Assessment of the patient with lithium toxicity

Whn one is asked to outline one's assessment of the patient, the goals should be broadly the same for all toxicology patients:

• Immediate risk assessment
• Looking for pattern features which might point to the lithium toxidrome
• Looking for causes of the intoxication
• Looking for sequelae

Thus:

History in acute and chronic lithium toxicity 

would typically be involving an intentional overdose for the acute version, or a gradual descent into chronic toxicity associated with any of the following risk factors:

• Recent infection
• A change in medication:
  • new NSAIDS
  • new ACE inhibitors
  • Thiazide diuretics
  • a new dose of lithium 

Presenting complaints are also different. Whereas, chronic toxicity has mostly CNS manifestations, of which the most unique are the cerebellar (others often being subsumed into the coma).

• Acute toxicity typically presents with mainly thoughts of suicide, GI upset (nausea, vomiting, bloating) and fluid-related features (polyuria and polydipsia) 
• CNS features (eg. confusion, falls) are dominant with chronic overdose.

Examination findings in lithium toxicity 

In acute toxicity, the only finding may be a tremor. CNS involvement, if it occurs, will be delayed because of slow distribution into the CNS. The only early findings are often hypovolemia, minor ST and T-wave changes on ECG, and a low urine output.

In chronic toxicity, the findings are mostly CNS. those listed in Okusa & Crystal (1994) include:

• Confusion progressing to coma
• Cerebellar symptoms:
  • Dysarthria
  • Ataxia
  • Nystagmus
  • Tremors
• Basal ganglia
  • Choreiform movements
  • Parkinsonian movements
• Seizures

The Hansen and Amdisen classification (Hansen & Amdisen, 1978) separates chronic CNS toxicity from litihium into risk categories:

• Mild: Tremor, hyperreflexia, agitation, muscle weakness, ataxia
• Moderate: Stupor, rigidity, hypertonia, hypotension
• Severe: Coma, seizures, myoclonus.

As this classification was devised on the basis of observing 23 overdose presentations, it does not seem to get a lot of airtime in the modern toxicology literature. 

Gastrointestinal features (nausea, vomiting, bloating) and fluid-related features (polyuria and polydipsia) are probably also present in chronic toxicity, but one needs to be awake and alert to give this history, whereas one commonly is not.

Investigations and biochemistry in lithium toxicity

Some broad sweeping statements can be made about lithium toxicity

• The anion gap will be negative  in acute toxicity, but probably low in chronic toxicity
• Bloods may be initially largely normal in acute toxicity, whereas in chronic toxicity there is often acute kidney injury, hypernatremia and hypothyroidism.

The greatest difference is in the serum levels, which can be made memorable by some silly mnemonic devices:

• In acute lithium toxicity, the levels are high but the patient is not.  Truly preposterous blood levels seem to be well tolerated, as the drug is slow to penetrate the brain. The patient may be minimally symptomatic with levels even above 4 mmol/L; peak levels > 5 mmol/L would not be unusual (to be clear, the normal therapeutic range for lithium is 0.8 and 1.2 mmol/L).  The level is therefore only performed once, to confirm the diagnosis.
•  In chronic lithium toxicity, the levels are as flat as the patient. As the drug has been accumulating chronically it has had abundant time to eqilibrate with all the body fluids and a mildly raised or normal level is meaningless. It will go up and down during management, and the main reason for repeat testing is to assess the need for further dialysis (as the plasma levels have absolutely no relationship to the clinical features). If one's repeated levels rise appreciably over the duration of the break between CRRT sessions, one needs to continue those sessions until it stops doing that. Which brings us to:

Management of acute lithium toxicity

To apply the usual RSI-DEAD mnemonic:

• Resuscitation

  • A - control the airway (the chronic lithium overdose may be sufficiently unconscious to be intubated).

  • B - maintain normoxia 

  • C - establish access resuscitate with fluids/vasopressors

  • D - Seizure control

  • E - Correction of hypothermia and hypernatremia, volume replacement after the polyuria of DI
  • There are no rescue antidotes for lithium 

• Decontamination is usually impossible with ingestion, as it is rapidly absorbed and in any case does not bind to activated charcoal.

  • Some report good results from cation exchange resin in animals, but the doses they used (2.5-5g/kg) are massive (equivalent to 350g in a 70kg patient!) and would surely cause more life-threatening electrolyte derangement than the toxicity itself.

• Enhanced elimination

  • Clearance is dependent on renal function and so efforts to restore the circulating volume will also restore clearance and help correct the toxicity.

  • Normal saline is the preferred volume replacement, as the sodium will be retained and lithium, hopefully, eliminated. There is more than just a theoretical basis for this; for example Lavonas & Brent (2017) report stories of crafty psychiatric inpatients consuming large amounts of kitchen salt to reduce their lithium levels and to thereby ensure that they remain mad.

  • Elimination is apparently also enhanced by theophylline and caffeine (Perry et al, 1984) but it is not clear whether this has much of a role to play in the acute management, considering the toxicity of these agents (i.e. does this patient need to be more agitated?)

Management of chronic lithium toxicity

The first part of RSI-DEAD will play out much the same way in all these cases. 

• Decontamination is even more impossible.
• Elimination by rehydration alone will not cut it. Dialysis is often called for. It is said that 2.5 mmol/L is the usual cutoff for considering dialysis in chronic toxicity, but realistically if the patient has modestly impaired renal function, then they cannot be expected to clear the drug themselves in a reasonable timeframe, and the impatient intensivist will usually connect them to the filter. On the other hand, the EXTRIP workgroup recommend 4.0 mmol/L as the cutoff. Considering that a room full of toxicologists could not agree on an accepted value, giving mostly random guesses when presented with case studies, one may be surprised that they formed any kind of workgroup at all. In general most people would ultimately agree that the role of dialysis in lithium toxicity is not clearly established, as the patient man can potentially ride it out without extracorporeal clearance, and the only limiting factor seems to be time and behaviour control.

Complications and known sequelae

Mortality from lithium toxicity is extremely rare and very few patients require dialysis unless the overdose is chronic and serious. Of the Ott cohort, n=1340, only twelve required CRRT. Long term neurological sequelae do occasionally follow, referred to with the terrifying acronym SILENT (syndrome of irreversible lithium-effectuated neurotoxicity). This is mostly cerebellar dysfunction, which is bad enough, but also akathisia dyskinesia and dementia are reported. 

Relevant guidelines and evidence

The best resource for this are the EXTRIP (EXtracorporeal TReatments In Poisoning) guidelines, which cover lithium. Beyond that, no better recommendations exist than the Toxicology Handbook.