Diagnostic accuracy of the anion gap to identify toxic lithium concentrations: a single-center retrospective observational study

Introduction: Lithium exhibits a narrow margin between therapeutic doses and toxic blood concentrations, which can pose a substantial risk of toxic effects. Reportedly, lithium toxicity may be associated with a reduced anion gap; however, the precise relationship remains unclear. This study examined several different anion gap calculation methods to detect toxic lithium concentrations without directly measuring blood lithium concentrations. Methods: Our retrospective study analyzed blood samples collected for lithium concentration measurements. The anion gap was determined using three different methods, both with and without albumin and lactate concentration corrections. Samples were categorized into two groups based on lithium concentration (<1.5 or >= 1.5 mmol/L), and anion gap values were compared. Correlation and logistic regression analyses were used to assess the relationship between each anion gap indicator and lithium concentration. Receiver operating characteristic curves were used for diagnostic analysis. Results: Overall, 24 measurements were collected, with 41.7% of samples falling within the toxic range. The high-lithium concentration group exhibited significantly smaller anion gaps. Correlation and logistic regression analyses revealed a significant association between anion gap values and lithium concentrations. Areas under the receiver operating characteristic curve were: conventional anion gap 0.77 (95% CI: 0.55-0.94); albumin-corrected anion gap 0.85 (95% CI: 0.66-1.00); and both albumin- and lactate-corrected anion gap 0.86 (95% CI: 0.66-1.00). Discussion: The anion gap is calculated as the difference between measured cations and anions. Accumulation of lithium (a cation) may decrease measured cations and decrease the calculated anion gap. Abnormal albumin and lactate concentrations may also alter the anion gap and affect its usefulness as a diagnostic marker for elevated serum lithium concentrations. A negative likelihood ratio of 0.1 suggests that the anion gap might be valuable in excluding toxicity. Conclusions: The corrected anion gap, accounting for albumin and lactate concentrations, may be beneficial in suggesting the possibility of toxic lithium concentrations.