To the Editor:
Plasma chloride is of value in the assessment of acid/base and electrolyte abnormalities. It is a component in calculation of both the anion gap and strong ion differences. The analytical method used extensively today, particularly in high-throughput analyzers, involves an ion-selective electrode incorporating a quaternary nitrogen compound. This has superseded amperometric titration with silver electrodes, which is not amenable to high-throughput or multianalyte instrumentation, and mercuric thiocyanate-based colorimetric methods, which require disposal of a toxic waste product.
In 1986 and 1991, two reports (1)(2) were published that documented marked positive interference by bicarbonate on plasma chloride analyses performed by Hitachi 705 and 717 analyzers. No further reports have been published since then. We suspected that similar problems might be occurring with the Dade Dimension RxL analyzer because of the high frequency of negative anion-gap calculations in samples from patients with metabolic alkalosis. Reviews of external quality assurance data for Dade Dimension RxL analyzers showed measured chloride concentrations in samples with above-normal bicarbonate that were higher than the values obtained with other instruments.
We evaluated the effect of increasing concentrations of bicarbonate on chloride measured by six different analyzers in two sets of experiments, using an aqueous and a plasma matrix. In the first set, a human lithium-heparin-plasma pool was prepared with a bicarbonate concentration of 10 mmol/L and a chloride concentration of 109 mmol/L. These values were the means of the results obtained on the six different analyzers. From this pool we prepared four other samples by adding various amounts of sodium bicarbonate (sodium bicarbonate, 8.4 g in 100 mL; Pharmacia & Upjohn), giving final bicarbonate concentrations ranging from 10 mmol/L (pH 7.49) to 44 mmol/L (pH 7.69). In the second set, we prepared a sodium chloride solution of 107 mmol/L by adding 0.619 g of sodium chloride to deionized water to a final volume of 100 mL. We then prepared eight samples with bicarbonate concentrations ranging up to 50 mmol/L by adding the same sodium bicarbonate solution used in the first set. The samples were then analyzed for chloride and bicarbonate on the six different analytical systems with the manufacturers’ standard methods. To demonstrate that any changes were not attributable to the pH changes on addition of bicarbonate, we prepared a pooled plasma to which sodium hydroxide was added, increasing the pH from 7.58 to 8.40. The measured chloride concentration remained constant at 105 mmol/L.
The results (Table 1⇓ ) show a clear linear increase in chloride with increasing bicarbonate in all except the Vitros analyzer. The effect is much greater in the Dade Dimension RxL than the others. The slopes of the other four analyzers were similar and were the same for the two specimen sets. This indicates that, in four of the analyzers, the apparent chloride will increase by ∼3 mmol/L if bicarbonate increases by 30 mmol/L and will increase by ∼10 mmol/L in the Dade Dimension RxL analyzer.
It is difficult to comment on the bicarbonate concentration at which the chloride value is unaffected because of the between-analyzer variability, but it would be expected to be in the region of the bicarbonate concentration of the calibrators. There is a fundamental difference in chloride methods between the affected analyzers and the Vitros. The Vitros has an ion-selective electrode based on a silver/silver chloride electrode, whereas the others use an electrode based on a quaternary nitrogen compound. These results clearly establish that the problem of bicarbonate interference in ion-selective chloride electrodes is still present in modern analyzers and is of a degree that dramatically distorts the quantitative evaluation of the acid/base balance. This could disguise the presence of a renal tubular acidosis in an acidotic patient or a coexistent unmeasured anion in a patient with metabolic alkalosis.
This problem with most current chloride methods is unacceptable and needs correction. An immediate corrective measure is possible by introduction of an on-board equation modifying the measured chloride value based on the linear relationship between the magnitude of interference and the bicarbonate concentration. However, the best corrective action would be for manufacturers to correct the problem present in quaternary ammonium-based ion-selective chloride methods.
We thank the following individuals for assaying the samples on their respective analytical systems: Kate Marshall and Gus Koerbin, Prince Charles Hospital, Queensland Health Pathology Service (Dade Dimension RxL); Peter Dulley and Brendan Sheehy, Ipswich Hospital, Queensland Health Pathology Service (Roche Integra); Barbara Mottram, Mater Laboratories Services (Ortho Diagnostics Vitros); and Phil Myers, Queensland Medical Laboratories, Gold Coast (Bayer Advia 1650).
- © 2004 The American Association for Clinical Chemistry