Introduction

Platelets normally release potassium when they are activated during clotting.^1–3 In vitro, this process can cause an artifactual increase in measured potassium concentrations in blood samples.^1,4,5,6,7 In blood samples from dogs, factors contributing to artifactual increases in in vitro potassium concentration include a high platelet count,⁴ blood sample storage of several hours’ duration prior to analysis,⁴ and hemolysis if the dog was of a breed with a high intraerythrocytic potassium concentration (eg, Akita,^8,9,10 Jindo,^10,11 Shar Pei,¹² Shiba Inu, and Tosa Inu). Whether such a laboratory artifact occurs in blood samples from cats and might be influenced by some of the aforementioned factors has been explored with 1 experiment.¹³ The clinical implications of this artifact could include misidentification of normokalemic cats as hyperkalemic, and failure to identify hypokalemic cats when an artifactual increase in potassium concentration raises the value to within the reference interval. The goal of the study reported here was to compare the potassium concentrations measured in heparinized feline blood samples (analysis of plasma concentration) with those measured in feline whole blood samples that had been allowed to clot (analysis of serum concentration) in accordance with protocols regarding handling and storage of samples that are submitted to our diagnostic laboratory.

Materials and Methods

The sample for this prospective comparative study consisted of cats brought for medical care to the Veterinary Teaching Hospital of the Atlantic Veterinary College, University of Prince Edward Island. Printed posters were displayed around the College and email messages were sent to the University community to make veterinarians, staff, students, and other cat owners aware of the study and its aims. A study sample of at least 40 cats was sought, based on previous reports.^4,13 Written, informed consent was obtained from all owners of cats included in the study. The study was approved by the institutional animal care committee (protocol No. 19-021).

To be included in the study, cats had to be ≥ 4 months old (based on a known date of birth or by estimation through physical examination) and considered likely to tolerate phlebotomy. Exclusion criteria were hemodynamic or respiratory instability that precluded phlebotomy; noncooperation of the cat in advance of or during physical examination or phlebotomy, as observed by one of the investigators or anticipated by the cat’s owner; platelet count > 600 × 10⁹ platelets/L; WBC count > 20 × 10⁹ cells/L; and inability to obtain a complete blood sample with a maximum of 3 phlebotomy attempts. Cats were not excluded from the study if an automated platelet count could not be performed as part of the CBC because of platelet clumping as long as the platelet count was estimated as normal on the basis of microscopic examination of a blood smear.

Each cat was restrained for routine phlebotomy performed by a veterinarian, veterinary technician, or veterinary student under the direct supervision of a veterinarian or veterinary technician. The hair over an external jugular or medial saphenous vein (or both) was parted, and a few drops of tap water (isopropyl alcohol was not used to eliminate the possibility of isopropyl alcohol–associated hemolysis¹⁴) were applied to moisten the overlying hair and skin. A 2- to 3-mL sample of blood was collected with a 3-mL syringe (Monoject; Covidien) and a 22-gauge (n = 38 cats) or 23-gauge (1 cat), 1-inch needle (Monoject; Covidien) or a 23-gauge, 3/4-inch butterfly catheter (Vacutainer Safety-Lok blood collection set; BD Canada; 2 cats). Care was taken to minimize the risk of hemolysis both during phlebotomy and with sample handling. Immediately on completion of blood sample collection, the needle or butterfly catheter was disconnected from the syringe and specimen tubes were uncapped. The blood sample was expelled directly from the syringe into 5 tubes as follows: first, 2 aliquots were placed in a no-additive plastic tube (Microtainer; BD Canada; to derive serum), then 2 aliquots were placed in a plastic tube containing lithium heparin and polymer gel (Microtainer plasma separation tube; BD Canada; to derive plasma), and a final aliquot was placed in a glass tube containing tripotassium EDTA (Monoject K³EDTA [0.06 mL/3 mL draw]; Covidien; for a CBC). All tubes were recapped immediately after the addition of the blood aliquot, and those with an additive were gently inverted 8 to 10 times to ensure proper mixing. One veterinary summer student or technician was responsible for overseeing each phlebotomy event and for prompt handling and transport of blood samples in person to the on-site laboratory. The sample in the tube containing K₃EDTA sample was processed routinely, and a CBC was performed the same day that the sample was submitted (XT 2000IV serial No. 13186; Sysmex Canada Inc). Samples in no-additive tubes were allowed to clot. For each cat, 1 clotted sample and 1 heparinized sample were kept at room temperature (approx 20 °C), centrifuged at 2,012 × g (Heralus Megafuge 40R; ThermoFisher Scientific) for 10 minutes, and analyzed for potassium concentration within 60 minutes after collection (designated the baseline concentration). The other serum and plasma samples from each cat were stored without centrifugation at 2 to 8 °C, centrifuged 20 to 28 hours after collection (designated the 24-hour concentration), and analyzed for potassium concentration.

Quality assurance was performed internally twice a day and externally biannually for one analyzer (XT 2000IV serial No. 13186; Sysmex Canada Inc) and internally once a day and externally once each quarter for the other analyzer (Cobas 6000-c501; Roche Diagnostics). The laboratory reference interval for potassium concentration used was that established for serum obtained from blood samples collected from approximately 50 cats presented to area veterinarians for routine wellness appointments. Those samples had each been allowed to clot and had been centrifuged within 30 minutes after collection; the serum then had been separated and placed in a no-additive tube and submitted to our laboratory for analysis.

Statistical analysis

For statistical analyses, Excel for Mac version 16.40 (Microsoft) and GraphPad Prism 8 version 8.4.3.686 (GraphPad Software) were used. Data were assessed for normality with the Shapiro-Wilk test. Means ± SD were calculated for normally distributed continuous variables. Medians, ranges, and interquartile (25th to 75th percentile) ranges (IQRs) were determined for nonnormally distributed continuous variables. Means and medians of the differences between paired results (serum and plasma potassium concentrations at a given time point) for individual cats were calculated, rather than the differences of whole-group medians or means. Comparisons of nonnormally distributed continuous variables between groups (serum and plasma baseline concentrations and serum and plasma 24-hour concentrations) were performed with Wilcoxon matched-pairs signed ranked tests and are presented as medians, ranges, and IQRs. Spearman rank order correlation coefficients were calculated to assess the strength of any correlations between the serum-plasma potassium concentration difference and platelet count, reticulocyte count, or reticulocyte percentage. If a cat’s age or body weight was not recorded, that variable was excluded from analysis for that cat. When a platelet count was reported as a range of 25 × 10⁹ platelets/L (300 × 10⁹ to 325 × 10⁹ platelets/L, 325 × 10⁹ to 350 × 10⁹ platelets/L, or 400 × 10⁹ to 425 × 10⁹ platelets/L), which our laboratory did on the basis of a technician’s estimate platelet count from the examined blood smear when platelets were flagged by the analyzer as being low and no clumping was noted, the mean of the 2 range limits was used in data analysis. A value of P < 0.05 was considered significant.

Results

Sixty cats were recruited over a 4-month period. Nineteen cats were excluded from the study because of noncooperation (n = 6), insufficient blood sample volume (5), high WBC count (2), high WBC and platelet counts (2), inability to collect the blood sample prior to a laboratory cutoff time that would allow for same-day analysis of baseline samples and initiation of storage of 24-hour samples (2), hemodynamic instability (1), and clotted blood sample (1). Thus, 41 cats were included in the study population. The reason for presentation of the cats to the hospital was elective ovariohysterectomy or castration (n = 24), wellness examination (7), recheck of known medical disorders (cholelithiasis, chronic kidney disease, hyperthyroidism, or idiopathic hypercalcemia [4]), preoperative evaluation for other elective surgery (3), congestive heart failure (1), dermatitis (1), and evaluation of an incidentally detected heart murmur (1). There were 18 males (7 of which were castrated) and 17 females (6 of which were spayed); for 6 cats, the sex was not recorded. There were 33 domestic shorthair cats, 1 domestic longhair cat, and 1 British shorthair cats; for 6 cats, the breed was not recorded. Age was recorded for 16 cats; the median age was 10.5 years (range, 9 months to 20 years; IQR, 2.25 to 12.1 years). Body weight was recorded for 35 cats; the mean ± SD weight was 4.45 ± 1.26 kg (range, 2.30 to 7.05 kg).

The median volume of blood collected from the 41 cats was 3 mL (range, 2 to 4 mL). The samples were collected by a summer veterinary student dedicated to this project and under the supervision of a licensed veterinary technician (n = 17 samples), by a licensed veterinary technician (15 samples), by a final-year veterinary student under the supervision of a licensed veterinary technician (6 samples), a veterinarian (1 sample), a student plus a veterinarian (1 sample), or a student plus a licensed veterinary technician (1 sample). For the 41 blood samples, the mean WBC count and Hct were 9.27 ± 3.65 × 10⁹ cells/L (range, 2.6 × 10⁹ to 17.3 × 10⁹ cells/L) and 30.9 ± 6.5% (range, 21.9% to 43.8%), respectively. The median reticulocyte count was 45.41 × 10⁹ reticulocytes/L (range, 12.61 × 10⁹ to 142.6 × 10⁹ reticulocytes/L; IQR, 33.26 × 10⁹ to 58.02 × 10⁹ reticulocytes/L), and the median reticulocyte percentage was 0.61% (range, 0.22% to 1.93%; IQR, 0.46% to 0.76%). No leukocyte or platelet morphologic abnormalities were reported. For 29 blood samples, the median platelet count was 444 × 10⁹ platelets/L (range, 312 × 10⁹ to 580 × 10⁹ platelets/L; IQR, 338 × 10⁹ to 494 × 10⁹ platelets/L). Hemolysis was identified in 0 of 41 baseline plasma samples, 3 of 41 baseline serum samples, 2 of the 41 plasma samples analyzed after 24-hour storage, and 6 of the 41 serum samples analyzed after 24-hour storage. Lipemia was identified in 1 baseline serum sample.

The median baseline serum potassium concentration (4.3 mmol/L [range, 3.4 to 7.4 mmol/L; IQR, 4.05 to 4.7 mmol/L]) was significantly (P < 0.001) higher than the median baseline plasma potassium concentration (4.1 mmol/L [range, 2.7 to 6.8 mmol/L; IQR, 3.6 to 4.25 mmol/L]; Figure 1). The median difference between baseline serum potassium concentration and baseline plasma potassium concentration was 0.4 mmol/L (range, –0.4 to 0.8 mmol/L; IQR, 0.1 to 0.55 mmol/L). Samples from 19 of 41 (46.3%) cats had baseline serum potassium concentrations > 0.4 mmol/L higher than the corresponding baseline plasma potassium concentrations. One of 41 (2.4%) cats had a baseline serum potassium concentration that exceeded the upper laboratory reference limit of 5.2 mmol/L (Cobas 6000-c501; Roche Diagnostics; Diagnostic Services, Atlantic Veterinary College, University of Prince Edward Island, Charlottetown, PE, Canada), although the baseline plasma potassium concentration was within the reference interval. In 1 cat, the baseline plasma potassium concentration was less than the lower reference limit of 3.3 mmol/L but the corresponding baseline serum potassium concentration was not (2.7 vs 3.4 mmol/L), indicating masked hypokalemia or pseudonormokalemia.^7,14

View Full Size

Potassium concentrations in serum and plasma samples obtained from 41 cats and analyzed within 60 minutes after collection (baseline) and after refrigerated storage for 20 to 28 hours (24-hour samples). Each box represents the interquartile (25th to 75th percentile) range, and the band across each box is the median. Whiskers represent the minimum and maximum values. The gray-shaded area indicates the laboratory’s reference interval for serum potassium concentration in cats.

Citation: Journal of the American Veterinary Medical Association 260, 2; 10.2460/javma.20.09.0530

• Download Figure
• Download figure as PowerPoint slide

After 20 to 28 hours of storage, serum samples from 16 of 41 (39.0%) cats had potassium concentrations that were > 0.4 mmol/L greater than the potassium concentration in the corresponding baseline serum sample. The median difference in potassium concentration between all 24-hour serum samples and all baseline serum samples was 0.4 mmol/L (range of differences, –0.3 to 1.0 mmol/L; 95% CI, 0.3 to 0.5 mmol/L; P < 0.001). After 20 to 28 hours of storage, plasma samples from 34 of 41 (82.9%) cats had potassium concentrations that were > 0.4 mmol/L greater than the potassium concentration in the corresponding baseline plasma sample. The median difference in potassium concentration between all 24-hour plasma samples and all baseline plasma samples was 0.6 mmol/L (range of differences, 0.1 to 1.5 mmol/L; 95% CI, 0.5 to 0.7 mmol/L; P < 0.001). After 20 to 28 hours of storage, serum samples from 38 (92.6%) cats had potassium concentrations that were > 0.4 mmol/L greater than the potassium concentration in the corresponding baseline plasma sample (range of differences, 0.4 to 1.4 mmol/L). The median potassium concentration in 24-hour serum samples was 0.1 mmol/L greater than the median potassium concentration in 24-hour plasma samples (range, –0.4 to 0.6 mmol/L; 95% CI, 0.1 to 0.2 mmol/L; P = 0.004). After 20 to 28 hours of storage, one or other sample from 7 of 41 (17.1%) cats had a potassium concentration that was outside the reference interval. Of those 7 cats, 5 had both serum and plasma potassium concentrations that exceeded the upper reference limit, 1 had a serum potassium concentration that exceeded the upper reference limit, and 1 had a plasma potassium concentration that exceeded the upper reference limit.

There was no significant correlation between reticulocyte count and the difference between serum and plasma potassium concentrations at baseline (r = 0.25; P = 0.11) or after 20 to 28 hours of storage (r = –0.017; P = 0.91). Similarly, there was no significant correlation between reticulocyte percentage and the difference between serum and plasma potassium concentrations at baseline (r = 0.23; P = 0.16) or after 20 to 28 hours of storage (r = 0.037; P = 0.82). There was no significant correlation between platelet count and the difference between serum and plasma potassium concentrations at baseline (r = 0.34; P = 0.07) or after 20 to 28 hours of storage (r = –0.048; P = 0.81).

Discussion

In mammals, potassium is the most abundant intracellular cation. However, there is considerable interspecies variation in intraerythrocytic potassium concentration. In cats, the intraerythrocytic potassium concentration (median concentration, 5.9 ± 1.9 mmol/L¹⁵) resembles the plasma potassium concentration because feline RBCs do not appear to have sufficiently effective transmembrane Na⁺-K⁺-ATPase pump activity to store K⁺ ions within erythrocytes against a concentration gradient.¹⁶ This is in contrast to human RBCs, which do have active transmembrane Na⁺-K⁺-ATPase pumps¹⁶ and a correspondingly much higher intraerythrocytic potassium concentration (median concentration, 102 ± 3.9 mmol/L).¹⁷ Transmembrane Na⁺-K⁺-ATPase activity also is responsible for the high intraerythrocytic potassium concentration in dogs of certain Asian breeds.¹¹ Such activity accounts for the artifactual hyperkalemia that occurs in association with hemolysis in blood samples from humans and from dogs of those breeds, but not with hemolysis in blood samples from cats or other breeds of dogs.

Although intraerythrocytic potassium concentration in adult dogs of non-Asian breeds and cats is similar to the circulating potassium concentration, intracellular potassium concentration is high in erythrocyte precursors in dogs and cats.¹⁸ Transmembrane Na⁺-K⁺-ATPase pumps exist in such precursors and normally are shed during maturation of precursors to erythrocytes.¹⁹ Thus, artifactual hyperkalemia can occur if there is hemolysis in blood samples from dogs of any breed when those dogs have substantial reticulocytosis. Similarly, erythrocyte precursors in cats have a high potassium concentration, which is decreased in mature erythrocytes.¹⁸ We are not aware of published evidence to prove whether reticulocytolysis causes an increase in measured potassium concentration in cats.

Platelets release K⁺ ions within minutes after the onset of clotting in blood samples.² In platelets, potassium exists mainly in the cytoplasm² with small amounts in the dense granules.²⁰ Human platelets have a potassium concentration of 86.4 mEq/L,²¹ which is approximately 20 times the plasma potassium concentration. As in human erythrocytes, human platelets maintain this concentration gradient by Na⁺-K⁺-ATPase pump activity. The same may be true for feline platelets given the results of the present study and a previous investigation,¹³ but we are not aware of published reports regarding the existence of such pumps in feline platelets. During clotting, platelet activation causes active K⁺ ion release; exposure to thrombin triggers a significant increase in measured potassium concentration.¹ Thus, it is logical to expect platelet-derived K⁺ ion release to contribute to artifactual hyperkalemia in vitro in baseline serum samples but not baseline plasma samples. With time, as platelets senesce in vitro and their supply of ATP is exhausted, Na⁺-K⁺-ATPase pumps cease to function and platelet K⁺ ions are released from platelets regardless of sample clotting. This mechanism would explain the higher potassium concentration in both serum and plasma samples after 20 to 28 hours of storage, compared with baseline findings in the present study.

Pseudohyperkalemia is defined as serum potassium concentrations > 0.4 mmol/L higher than the corresponding plasma potassium concentration.^22,23 A fictitious increase in measured potassium concentration can be caused by physiologic phenomena involving blood cells (eg, hemolysis in blood samples from humans and specific aforementioned breeds of dogs, WBC lysis, and platelet activation), patient factors during blood collection (eg, the patient clenching the fist²⁴ or squeezing a rubber ball²⁵ during phlebotomy), and technical errors^7,14,23,26 (eg, contamination of the needle tip when a portion of a blood sample is injected into a tube containing K₃EDTA before a portion is injected into a no-additive or heparinized tube and hemolysis induced by aspiration of isopropyl alcohol placed on the skin of the phlebotomy site).

In the present study, cats underwent routine phlebotomy. Nevertheless, the measured potassium concentration in approximately half of the baseline serum samples differed from the corresponding baseline plasma potassium concentrations by > 0.4 mmol/L; moreover, the potassium concentration in 82.9% of plasma samples and 92.6% of serum samples after 20 to 28 hours of storage differed from the corresponding baseline potassium concentrations by > 0.4 mmol/L. These differences suggest that tube selection and sample storage, respectively, can alter a large proportion of feline blood samples submitted for potassium concentration determination. Importantly, the wide range of differences indicated that the degree of artifactual hyperkalemia cannot be predicted reliably and therefore cannot be corrected for. Nevertheless, samples submitted for measurement of potassium concentration in cats often are serum from clotted blood in no-additive tubes. For example, in 2019, our Diagnostic Laboratory received and processed 1,718 feline blood samples for analysis of electrolyte concentrations, and the majority (1,577 [91.8%]) were serum; of those 1,718 samples, 907 (52.7%) were without a blood clot and 670 (40.0%) contained clotted whole blood. Only 141 (8.2%) of the submissions were plasma samples. Of the 1,718 samples, 72 (4.2%) plasma samples were without cells and 69 (4.0%) plasma samples contained heparinized whole blood (data on file; Diagnostic Services, Atlantic Veterinary College, University of Prince Edward Island, Charlottetown, PE, Canada). Results of the present study have provided justification for choosing to submit samples of plasma obtained from heparinized blood samples rather than choosing to submit samples of serum. These results also provided justification for centrifuging plasma samples and separating plasma from cells in the samples to avoid artifactual increases in potassium concentration of an unpredictable magnitude. This recommendation applies even to samples processed within 60 minutes after collection because significant differences between plasma potassium concentration and serum potassium concentration develop during this time.² Similarly, the results of the present study supported a recommendation that feline plasma–specific and feline serum–specific laboratory reference intervals for potassium concentration be derived and the appropriate reference interval be used depending on the type of sample submitted.

The present study did not identify a significant relationship between platelet count and the difference between feline plasma and serum potassium concentrations. Such a relationship exists in dogs⁴ and people, although it is most pronounced in individuals with thrombocytosis.^27,28,29 The lack of a demonstrable relationship between platelet count and potassium concentration could have been attributable to a number of variables, including lack of thrombocytosis in most of the cats in this study (although it is recognized that platelet clumping is frequently found in blood samples collected from cats^30,31 and that the results of the present study could have underrepresented true platelet counts), feline-specific interspecies differences including differences in platelet function,^26,32 or underpowered statistical analyses.

In the present study, the passage of time between feline blood sample collection and analysis was associated with an increase in potassium concentration in both serum and plasma samples. The serum or plasma potassium concentration (or both) in samples analyzed at 20 to 28 hours after collection exceeded the upper reference limit at our laboratory for 7 of 41 (17.1%) cats. Such a result could prompt unnecessary investigation for disorders that cause hyperkalemia. However, it should be noted that the reference interval used at our laboratory was determined from serum samples that had been centrifuged and separated within 30 minutes after blood sample collection. The strength of this observation should be tempered by the possibility that a plasma-specific reference interval for potassium concentration might differ from that used for assessments in this study, although a plasma-specific reference interval likely would be comparatively lower, thereby increasing rather than decreasing the number of false-positive findings after 20 to 28 hours of storage.

Earlier work used a similar approach to investigate artifactual increases in serum potassium concentration in cats.¹³ That pioneering study¹³ was the first to identify this phenomenon in cats. The present study sought to test the general reproducibility of those results, while bringing changes to the protocol that could have important effects on results. Specifically, we used a 20- to 28-hour intermeasurement interval instead of 48 hours to more closely mimic the time between sample collection and analysis in our referral population, the samples were refrigerated for the 20- to 28-hour intermeasurement period instead of kept at room temperature, and the time between blood sample collection and baseline analysis of samples was shorter than the currently recommended time²⁶ (< 1 hour). Moreover, we compared our results to existing reference intervals to determine how practitioners might interpret their patients’ results, additional variables (reticulocytosis and hemolysis) were evaluated, and we standardized blood sample collection and handling techniques to minimize collection- and processing-related artifacts. Whether the differences in study results were attributable to our modifications, other factors, or merely chance, the present study revealed quantifiable platelet counts in 30 of 41 (73.1%) cats (all of which were within or exceeded the reference interval), whereas the previous study¹³ revealed quantifiable platelet counts in 17 of 41 cats, and only 7 (17.1%) were within or exceeded the reference interval. The results of the present study indicated that under different conditions of sample acquisition, handling, and storage, significant artifactual changes in potassium concentration still occur.

Historically, blood potassium concentration in cats has been determined only in serum samples^33–36 or without distinguishing between serum and plasma samples.³⁷ Therefore, artifactual increases in measured potassium concentration may be widespread and could have variable effects of unspecified magnitude. Repercussions of inaccurate measurement of potassium concentration in feline serum samples could include under-recognition of hypokalemia in predisposed populations,^35–37 misleading assessments of responses to potassium supplementation,³⁷ disagreement between the apparent evidence of hyperkalemia and other clinical findings,³⁴ and errors in sodium-potassium concentration ratio calculations.³³

It is possible for a low blood potassium concentration to appear to be within the reference interval if the sample that is analyzed is serum instead of plasma, thereby causing hypokalemia to be missed. This phenomenon is called masked hypokalemia or pseudonormokalemia.^7,25 In the present study, it was uncommon, with only 1 of 41 (2.5%) cats affected. However, there was only 1 cat with hypokalemia in the study population. Both type I and II errors are possible when considering hypokalemia and in vitro potassium concentration changes; hypokalemia could be underrecognized because measurements are made in serum samples rather than in plasma samples, and the prevalence of hypokalemia could be overstated if plasma samples are assessed against reference intervals determined from serum samples.

The difference between statistical significance and clinical relevance of data was an important consideration in the present study. Although a significant difference was noted between potassium concentrations in serum and plasma samples and between potassium concentrations in baseline samples and samples stored for 20 to 28 hours, most potassium concentrations remained within the reference interval. This observation could lessen the importance of artifactual differences in potassium concentration in cats caused by in vitro effects. However, the cats in the present study were not selected for illnesses that predispose to hypo- or hyperkalemia, and cats with spontaneously occurring leukocytosis or thrombocytosis were not included in the study, even though such hematologic abnormalities are known to be associated with pseudohyperkalemia in other species.^1,4 The clinical relevance of the present results may be most important in cats with a circulating potassium concentration that is near either limit of the reference interval.

Limitations of the present study included the number of cats enrolled (which precluded subgroup analyses), lack of demographic information for 6 of the 41 cats, inclusion mostly of cats with apparently normal baseline potassium concentration, and lack of separate reference intervals for plasma potassium concentration and serum potassium concentration. Further investigations in which these limitations are overcome are warranted.

As reported previously for cats, humans, and dogs,^1,4,13 potassium concentration measured in feline serum samples was higher than that measured in plasma samples when analyses were conducted prospectively under controlled conditions of blood sample collection and handling. Although many of the study cats had potassium concentrations that were within the reference interval, there is the potential for in vitro K⁺ release to have clinically important implications for patients in which K⁺ disturbances may be anticipated. Prompt measurement of potassium concentration in plasma samples and centrifugation of blood samples and removal of supernatant from cells (or both) may be considered to avoid this artifact.