Abstract
Blood samples were collected from 41 cats presented for pre-anaesthetic assessments, routine geriatric screening, or re-assessment of ongoing chronic medical disorders. Samples were either left to clot or anticoagulated with lithium heparin, then assessed for their potassium concentration within 1 h of collection, and again after remaining in contact with their cell pellet for 48 h. There was a significantly higher potassium concentration in the serum samples compared to the plasma samples, both in the basal and 48-h samples (although this difference was most marked in the basal samples). Ageing of both serum and plasma samples also resulted in an increase in the potassium concentration when compared with the basal values for each sample type. The mean difference (basal serum minus basal plasma) in potassium concentration was 0.47 mmol/l. While it is probable that the potassium came from either leukocytes and/or thrombocytes the mean total leukocyte count and the mean thrombocyte count were below the upper limit of the reference intervals for our laboratory and the rise in the potassium level did not appear to be directly related to either of these values.
The significance of a false elevation in potassium concentration will vary depending on the true potassium concentration. Therefore, while it may not result in a change that takes the patient's result outside the reference interval, it can also result in pseudohyperkalaemia (where the potassium concentration is raised above the upper end of the reference interval) or pseudonormokalaemia (where a hypokalaemic sample is raised to within the normal reference interval).
As potassium is released from its intracellular location in response to blood clotting, a false elevation in potassium concentration is typically seen in serum samples, rather than plasma samples. This is a common finding in samples from human beings, where pseudohyperkalaemia is classically diagnosed when the serum potassium concentration exceeds the plasma potassium concentration by 0.4 mmol/l (Singh et al 1997). While this definition does not take the value of the reference interval into account, the statement does indicate the degree of change that is likely to be significant. Until recently there was little evidence of the importance of this phenomenon in cats and dogs. However, it has now been shown that dog serum typically has a potassium concentration 0.35 mmol/l higher than that of matched heparin plasma (Cerón et al 2004). It has also been suggested that a similar situation may occur in cats (Kim Kendall and Randolph Baral, personal communication, 2005), however, to the authors' knowledge, no studies have yet been published. The aims of this study were, therefore, to determine whether a false increase in potassium concentration occurs in feline serum when compared to matched heparin plasma samples, and to assess the effect of delayed separation of the cell pellet on plasma and serum potassium concentrations.
Materials and methods
Blood samples were collected from 41 cats presented to the Royal (Dick) School of Veterinary Studies (R(D)SVS) Hospital for Small Animals for pre-anaesthetic assessments, routine geriatric screening, or re-assessment of ongoing chronic medical disorders. Samples were collected between September 2003 and February 2005. The cats had a mean age of 9.4 years (range 2–19 years); 17 were females, 24 were males, and all were neutered. No breeds were over-represented; 30 were domestic short haired cats, while 11 were pedigree, with eight different breeds being represented. Serum samples were also collected from 17 clinically normal cats to establish a contemporaneous reference interval for cats from the same population.
Blood samples were collected via jugular venepuncture and then divided between a tube containing lithium heparin as anticoagulant, a tube without anticoagulant, and a tube containing potassium EDTA as anticoagulant and submitted to the laboratory within 1 h of collection. The blood in the non-anticoagulated tube was allowed to clot for less than 2 h at room temperature (approximately 17°C), then centrifuged at 1250 g for 5 min and an aliquot of serum was removed for immediate analysis (basal serum). The remainder of the clotted sample was left in the tube in contact with the clot for 48 h at room temperature after which the serum was again analysed for potassium (48-h serum). The contents of the heparinised tube were divided into two aliquots; one was centrifuged and the plasma harvested for immediate analysis (basal plasma). The second was left at room temperature for 48 h, then centrifuged and the plasma analysed (48-h plasma). Routine haematology was performed using EDTA anticoagulated blood. Blood samples were collected into tubes without anticoagulant from the 17 normal cohort cats and the serum harvested for baseline potassium measurements as described for the study population. Heparinised plasma was not collected from these cats, nor was the serum aged.
All biochemistry analyses were carried out by indirect potentiometry on an AVL 9180 Electrolyte Analyzer (AVL Scientific Corporation, Roswell, GA, USA) using an ion-specific electrode. Routine haematology was performed on a Pentra 60 haematology analyser (ABX Diagnostics, Montpellier, France), and the manual thrombocyte count was determined in a haemocytometer slide using standard methods, and only included where clumping was minimal. The reference intervals had been previously defined for this laboratory, using EDTA anticoagulated blood and serum samples from 50 normal cats. However, they were rechecked by running the serum samples from 17 healthy cats in parallel to the samples studied in this paper.
Statistical analyses
The paired values of ‘basal serum’ versus ‘basal plasma’, ‘48-h serum’ versus ‘48-h plasma’, ‘basal serum’ versus ‘48-h serum’ and ‘basal plasma’ versus ‘48-h plasma’ were compared using paired t-tests. An Altman–Bland difference plot was used to assess the difference between basal serum and basal plasma potassium concentrations by comparing the mean of both methods. In four animals, there was insufficient serum or plasma for the 48-h potassium estimation, in one animal there was no 48-h sample, and in two animals, only a heparinised sample was available.
Results
The results obtained for each sample are summarised in Table 1 and the frequency distribution is shown in Fig. 1. There was a significantly higher potassium concentration in serum than plasma, both in the basal and 48-h samples, with the difference being most marked in the basal samples. Ageing of both serum and plasma samples also resulted in an increase in the potassium concentration when compared with the basal values for each sample type. The basal serum potassium concentration was consistently higher than the basal plasma potassium concentration (Fig. 2) with the exception of 2/39 data points in which basal serum potassium was lower than plasma and 3/39 in which basal serum and plasma potassium were equal. The mean difference (basal serum minus basal plasma) in potassium concentration was 0.47 mmol/l.
Frequency of potassium concentrations for each sample treatment category. Red vertical dotted lines indicate R(D)SVS reference interval and green vertical dotted lines the reference interval for normal cats (n=17) from the same population as the study cats.
Altman–Bland difference plot to assess the difference between basal serum and basal plasma potassium concentrations compared with the mean of both methods. There is a marked negative bias associated with the use of plasma compared with serum.
Basal and 48-h potassium results in heparinised plasma and serum, plus total white cell counts and manual thrombocyte counts
The normal cats were sampled from the same clinic population at the same time as the study cats. Of the comparisons made, parameters with the same superscript are significantly different; a, c and d: P<0.001, b: P=0.002.
Manual thrombocyte count with no minimal clumping.
Median=9.2×109/l.
Leukocyte counts were available for 34 of the cats (Table 1). Only three of the 34 samples had total leukocyte counts outside the reference interval; their values being 24.7, 34.0 and 118.4×109/l, which corresponded to basal plasma (serum) potassium concentrations of 2.9 (2.9), 4.0 (4.5) and 3.4 (4.2) mmol/l, and post-48-h plasma (serum) potassium concentrations of 3.5 (3.3), 4.5 (5.1) and 6.1 (5.8) mmol/l, respectively.
Manual thrombocyte counts (without significant clumping) were available for 17 of the cats (Table 1). None of the cats had thrombocytosis although 10/17 were thrombocytopenic (ie, less than 200×109 thrombocytes/l).
Discussion
This study shows that when analysing feline blood, the use of serum is typically associated with a higher potassium concentration than matched heparin plasma. In addition, delayed separation of the serum or plasma from the cell pellet will also lead to an increase in potassium concentration. The degree of the increase was greater than has been reported for dogs, with the mean difference (basal serum minus basal plasma) in potassium concentration being 0.47 mmol/l compared to 0.35 mmol/l with dog blood (Cerón et al 2004). This is in close agreement with unpublished data from 16 samples of feline blood analysed using the IDEXX Vetlyte, where the mean difference (serum minus plasma) in potassium concentration was 0.6 mmol/l (Kim Kendall and Randolph Baral, personal communication, 2005).
The basal serum samples gave higher potassium concentrations than the basal plasma samples which is consistent with the potassium having been released during clot formation. While it is probable that the potassium came from either leukocytes and/or thrombocytes the mean total leukocyte count and the mean thrombocyte count were well below the upper limit of the reference intervals for our laboratory. Only three of the 34 samples had total leukocyte counts outside the reference interval; there was no clear relationship between leukocytosis and the size of the increase in potassium concentration but interestingly, the cat with the highest leukocyte count did have the highest post-48-h plasma and serum potassium concentrations of all the cats in the study. Therefore, while it is possible that the potassium leaks from leukocytes during the clotting process, it appears that significant leukocytosis is not necessary for a significant increase in potassium concentration to occur.
The possibility that the potassium is released from thrombocytes was considered because it has previously been suggested that thrombocyte counts of greater than 1000×109/l (reference interval 200–400×109/l) may result in pseudohyperkalaemia in feline blood samples (Loar 2002). In human medicine, where this phenomenon has been most closely studied, a thrombocyte count of 500×109/l is known to result in an increase in the serum potassium concentration of approximately 0.5 mmol/l (compared to that of matched heparin plasma samples) and when the difference (serum minus plasma) in potassium concentration is plotted against thrombocyte count, there is a direct linear relationship (Thurlow et al 2005). It is because of this, where blood potassium concentrations are found to be greater than 5.4 mmol/l, it is recommended that the thrombocyte count be checked before interpretation in human beings. Interestingly, in the current study, none of the samples had thrombocyte counts above the upper limit of the reference interval for our laboratory. While this may suggest that in cats thrombocytosis is not necessary for a significant increase in potassium to occur, it is important to remember that cat thrombocytes are very susceptible to clumping, so it is very difficult to say exactly what the true thrombocyte count was in any of the samples tested. It should also be noted that a high percentage of these samples appeared to be thrombocytopenic. The reason for this is unclear as few of the cats were particularly ill, it may instead relate to samples being collected by veterinary students, where thrombocyte activation and clumping may have occurred before the sample entered the EDTA anticoagulant, which could result in a spurious thrombocytopenia, without obvious clumping being seen on the slide.
This study shows that it is important to consider whether serum or plasma has been used before determining the true significance of a particular potassium concentration in feline blood. In addition, it is important to know whether the reference interval for a particular laboratory has been generated using serum and/or plasma samples, and whether or not the reference samples were processed promptly. It could be argued that as most of our serum results were within our reference interval these results are likely to accurately reflect the cats' true potassium status. However, this misses the important point that the serum results may be artificially and variably increased by the leakage of potassium from the blood cells. It is, therefore, preferable to use heparin plasma for accurately assessing the potassium concentration in cats. It is also important that the samples are either processed quickly, or separated from the cell pellet prior to posting to a diagnostic laboratory. In addition, diagnostic laboratories, including our own, need to generate a separate reference interval for plasma samples. Even small differences in blood potassium concentration can lead to significant clinical consequences the accurate measurement of this cation is essential if we are to ensure that appropriate treatments and/or interventions are undertaken.
Footnotes
Acknowledgements
The authors wish to thank all members of the R(D)SVS staff who helped in this study. In addition, KS's and NR's positions were supported by the Feline Advisory Bureau.