Effects of Hemolysis on the Roche Ammonia Method for Hitachi Analyzers

The manufacturer’s product literature for the Roche ammonia reagent (cat. no. 11877984) states that “there is no interference from hemolysis up to a hemolysis index of 50 (approximate hemoglobin concentration: 50 mg/dL)”. This statement can and has been interpreted to mean there are artifactual changes in the ammonia results above this hemolysis index limit. The likely result is that laboratories could use this low limit as the definitive guide and not report any results above it. This would lead to rejection of up to 10% of samples. The effect of hemolysis on analytes is often studied [e.g., Ref. (1)] by adding a hemolysate(2) to a pool of serum/plasma. To investigate the effects of hemolysis on ammonia, we used a slightly different method. We split an ∼50-mL freshly collected EDTA-blood sample (recommended sample type) from a volunteer into three pools. Pool 1 was immediately aliquoted into equal sample volumes (2 mL). The first sample was centrifuged, and the EDTA plasma that was separated contained minimal hemolysis. The remaining samples were passed through a blood collection needle to mimic a typical collection process. The number of times a sample was passed through a needle increased with each subsequent sample to produce a range of hemolysis. Ammonia was added to pools 2 and 3, which were then processed similarly to pool 1. The samples were centrifuged, and the EDTA-plasma was separated and analyzed on an Hitachi Modular P analyzer (Roche Diagnostics Australia). A separate EDTA-whole-blood sample was used to prepare hemolysate (2). The hemolysate was then added to portions of EDTA plasma to produce a range of hemoglobin concentrations. When hemolysis was induced by mimicking a typical collection process, hemoglobin up to 11 g/L had no clinically significant effect on the ammonia result (Fig. 1 ). By contrast, when the samples were prepared by adding hemolysate, there was a dramatic effect on the ammonia result. The most likely explanations for such increases in ammonia with prepared hemolysates are freezing and aging, which lead to increased ammonia production via deamidation of the cellular proteins in the hemolysate. We suspect that the statement in the Roche package insert was based on experiments in which hemolysate was added to plasma. Therefore, the effect on ammonia is most likely a preanalytical artifact rather than attributable to analytical interference from hemolysis. The data clearly point out the importance of mimicking actual clinical settings as closely as possible when performing interference studies. Effect of hemoglobin concentration on ammonium measurements. X, sample to which hemolysate was added; ▴, ▪, and ♦, samples mimicking a typical sample collection process. We observed no such deamidation of proteins in hemolyzed plasma over a 4-h time period, but to prevent any possible ammonia production as a result of deamidation of the cellular proteins, ammonia samples should be centrifuged immediately and the EDTA plasma separated from cells, sealed, and stored at 4–8 °C, not frozen.