Alert: Inaccurate Lithium Assay Results

Abstract

Objectives: A patient who had experienced bipolar disorder for over 20 years and who had been euthymic for most of that period while highly compliant with lithium, had falsely low lithium levels reported over two periods, 6 years apart, and was actually lithium toxic on the most recent occasion. At that latter time the spuriously low lithium levels reported on the assay risked dispelling any clinical suspicion of lithium toxicity, although toxicity was later confirmed.

Method: Case report.

Results: The latter incident identified a serious problem, whereby it is likely that the particular assay risks generating spuriously low values when high serum levels of lithium are present – a so-called ‘hook phenomenon’ that has been described for some quantitative immunoassays.

Conclusions: It is in situations of high potential gravity – when lithium toxicity is present – that lithium quantification is most likely to be compromised and low values generated. Also of grave concern is the fact that there are no regulatory processes in place to communicate this problem to lithium prescribers.

The wife of a 60-year-old professional man sought urgent assessment of her husband in March 2008. He had been initially referred 20 years previously following a severe episode of psychotic depression that had required hospital admission. He was then well for 6 months before he developed his first clear-cut manic episode, although he had experienced some milder hypomanic episodes over the preceding few years. There was a family history of bipolar disorder in two first-degree relatives. Following that initial admission, he was discharged from hospital on lithium, together with two adjunctive medications that were able to be ceased over the next few months. It was recommended that he stay on lithium carbonate (1000 mg day⁻¹) and, once stabilized, for review to be principally undertaken by his general practitioner.

At personal review 2 years later he had remained euthymic over the period, and had been totally compliant with his lithium monotherapy. His serum levels were fairly consistently around 0.8 mmol L⁻¹. Several years later, and following a surgical procedure resulting in some disruption to his lithium medication, he developed an acute manic episode that required hospitalization for nearly 2 weeks. Over the next decade he had remained generaly euthymic on lithium, with very occasional adjunctive medication, and, on a lithium carbonate dose of 1000–1250 mg day⁻¹, he returned relatively consistent 0.8 mmol L⁻¹ levels. Following the patient's diagnosis of hypertension in 2000, introduction of the antihypertensive drug lisinopril dehydrate (Privinil, lisinopril dihydrate) effectively increased his concentration of lithium and required a reduction of lithium carbonate to 625 mg day⁻¹ to maintain the 0.8 mmol L⁻¹ serum lithium level.

Subsequently, on two occasions in 2002 (September and October), he returned unusually low lithium levels of 0.4 mmol L⁻¹, following two high level of 1.9 mmol L⁻¹ in May and which had resulted in his lithium does being lowered. His wife queried the pathology company. The company re-checked the latter analyses, stating that there had been ‘clots in the tube’, and – on re-analysis – reported corrected levels of 0.8 mmol L⁻¹. It is worth noting that this may have been around the period that the automated lithium assay was introduced by that company.

In 2003, again following a surgical procedure, he developed mood elevation, which was judged as possibly reflecting a change in his lithium dose at the time of that operation. An antipsychotic medication was added temporarily, and his mood settled.

In essence, lithium had been a superb drug for this man for more than 20 years. Prior to its introduction he had experienced severe psychotic depressive and manic episodes. Following its introduction he had had few recurrences, had rarely required additional medication, had been largely able to be monitored by his general practitioner (with reviews by myself stretching up to 5 year intervals), had worked capably, and been a respected family and community member. Such a successful outcome was helped by his treatment compliance and his wife's vigilance with his lithium medication.

In mid- and late-2007 he had two surgical operations with less than satisfactory results and developed some stress and depressive symptoms. Over that time he had been receiving 1125 mg day⁻¹ of lithium with levels of approximately 0.8 mmol L⁻¹ (that level last recorded in a November 2007 assay), as well as 20 mg day⁻¹ of Prinivil. On 19 March 2008 his wife (a former nurse) organized an urgent psychiatrist review in response to her husband's gathering agitation and confusion, and possible acute depression. Being concerned about the possibility of lithium toxicity, she had initiated lithium level checks on 6, 11 and 18 March from a major Australian pathology company. On each of those occasions the lithium level was formally reported as 0.4 mmol L⁻¹. Other tests identified mild anaemia, slight neutropenia and a slightly elevated creatinine level, at 125 umol L⁻¹ (normal range 40–120 umol L⁻¹). There was also elevation of some liver enzymes (with his alanine aminotransferase level of 51 U/L exceeding the upper limit of 45 U/L and the γ-glutamyl transferase level of 121 U/L exceeding the upper limit of 121 U/L).

When I interviewed her husband on 19 March he did not appear clearly depressed, but was distinctly tremulous, pale and confused, and failed a number of tests of concentration and memory. While the picture was compatible with lithium toxicity, the three lithium levels of 0.4 mmol L⁻¹ in the preceding 2 weeks seemed to exclude that possibility. I recommended that he be hospitalized to pursue possible physical causes of his current state.

Following the patient's admission, his managing neurologist ordered an electroencephalogram that showed a pattern consistent with lithium toxicity. His lithium was suspended, and his lithium level was independently checked. Some 60 h after suspension of that drug, the hospital's pathology laboratory quantified a lithium level of 1.1 mmol L⁻¹. Following saline perfusion, lithium was slowly reintroduced and, after 3 days, his confusion had settled.

His wife requested that the pathology company investigate why three levels of 0.4 mmol L⁻¹ had been quantified over the previous 2 week period, when clearly his lithium level was much higher on each occasion. The company re-tested the last of his three blood samples, using a manual dilution strategy rather than relying on their customary assay procedure, and quantified a lithium level of 1.8 mmol L⁻¹. On further testing, they established that when a serum level was high, the assay (provided by an independent major pharmaceutical company) performs an ‘automatic dilution’ that can then generate a falsely low value. Thus, patients with high lithium levels and at risk of toxicity are, paradoxically, most likely to generate a spuriously low (and false) value, and therefore be at risk of inappropriate clinical management.

Such a phenomenon has been widely described in other domains. An early report by Goodman and Masaitis described the ‘prozone phenomenon’ in haemagglutination reactions, where there was a curvilinear relationship between quantitative agglutination values and an increase in antiserum concentration [1]. Now more commonly referred to as the ‘hook effect’, the phenomenon is one where, in quantitative immunoassays, increasing concentrations produce values that first rise and then fall.

Approximately a decade ago I had a series of patients return distinctly lower lithium levels than usual (although their lithium dose had been constant) but, on querying whether there had been a change in analytic procedures, I was assured that any such possibility of false quantification was unlikely. On this most recent occasion, when queried, the pathology company identified and admitted to an assay problem. When questioned about wider reporting, they stated that it would be regarded by the company as a ‘critical incident’ and that they would, of necessity, inform their chief scientist and the pharmaceutical company that provided the assay. But no other reporting mechanism was required because there is no regulatory authority that requires notification.

If there was a regulatory process for assay problems and anomalies, several questions might fruitfully be asked. For example, at what elevated level of lithium level does the risk of a false low lithium level emerge? Is that level a constant, or is the phenomenon restricted to a subset of patients, and, if so, what might characterize the subset? Is the problem idiosyncratic to the assay provided by one pharmaceutical company, or relevant to other assays? Does the problem lie with the assay or its calibration?

In the absence of any formalized regulatory mechanism for bringing this issue to prescribers, this report has been prepared. It is unlikely that this is a one-off event. It is quite extraordinary that one individual has experienced two periods of spuriously low lithium levels (in 2002 and 2008) for two seemingly differing reasons (i.e. ‘clot’ in collection tube and assay abnormality), although the possibility exists that the first event had a similar cause. And it was only as a consequence of assiduous checking by his wife of the possibility of faulty measurement processes that the anomalies were identified. When retesting occurred, quite differing values were returned on the separate occasions.

If one individual has experienced two differing procedural factors compromising valid quantification of a lithium level (and with the latter, in effect, allaying clinical suspicion of lithium toxicity), one may well query the prevalence of this problem. Given that Wills et al. noted a reporting rate of ‘lithium poisonings’ to US monitoring centres of more than 5000 year⁻¹ [2], we can conclude that lithium toxicity is not a rare phenomenon, and that the need for accurate quantification is vital. Any assay irregularities need to be identified, reported and corrected.

Until the current assay limitation has been clarified, the capacity of managing psychiatrists and other practitioners to be confident about lithium assaying is badly compromised. For this patient the company has agreed to use a manual dilution strategy in future rather than relying on the automated assays. But although his wife's and neurologist's vigilance have ensured a good outcome for him, his case serves to raise questions about the responsibility of pathology companies to report (and, indeed, to detect) such irregularities.