Preventing pseudohyponatremia: Intralipid®-based lipemia cutoffs for sodium are inappropriate

Background Pseudohyponatremia describes an artifactual decrease in plasma sodium result in samples with high proteins and/or lipids when measured by an indirect ion-selective electrode (ISE) method. We suspected that Intralipid®-based lipemia cutoffs are inappropriate for detecting interfering lipids in human samples and a major contributing factor to the existence of pseudohyponatremia. Methods We evaluated 2 approaches to derive a lipemia cutoff for sodium, one in which patient plasma samples were pooled and spiked to simulate hyperlipidemia using Intralipid® (commonly used approach by in-vitro diagnostics manufacturers), and another in which endogenous hyperlipidemic samples (n = 31) were measured by methods not affected by hyperlipidemia (i.e., direct ISE and post-ultracentrifugation indirect ISE). Triglycerides, lipemic index (L-index) and indirect ISE sodium concentrations of samples were measured on Roche Cobas® 8000 and direct ISE on Radiometer® ABL835 Flex analyzers. Endogenous hyperlipidemic samples were also ultracentrifuged on Beckman Coulter® Airfuge to clear excess lipids and re-analyzed for sodium by indirect ISE. Results We discovered that Intralipid® is not an accurate emulation of the lipemic interference seen in pseudohyponatremia because it showed no effect up to the maximum level of lipemia tested (L-index = 2000). By contrast, endogenous hyperlipidemic samples demonstrated significant deviations in sodium concentration (≥4 mmol/l) when L-index approached or exceeded 700, and a strong positive correlation between L-index and the difference between the indirect and direct methods (i.e., extent of pseudohyponatremia). Conclusions Clinical laboratories should lower their tolerance for lipemia from the currently recommended L-index cutoff of 2000 on Roche Cobas 8000®. We recommend reflexing to direct ISE when L-index exceeds 700. Manufacturers and laboratories with other indirect ISE methods should evaluate the effect of lipid interference on their method using hyperlipidemic human samples not Intralipid®.