Comparison of Two Portable Lactate Meters in Dogs

Introduction

Measurement of plasma lactate levels in the critical care setting is a well-established technique in human medicine both for evaluating response to treatment and for prognostic purposes.^1,2 The use of cage-side lactate analyzers is becoming more popular in veterinary intensive care units.

Lactate has been used clinically for prognostic purposes as well as trending response to treatment in states of either local or systemic tissue hypoxia in humans and dogs.^1–7 In normal states, oxygen delivery is more than adequate for tissue demands, and lactate utilization is independent of oxygen delivery to the tissues. As lactate utilization becomes more dependent on oxygen delivery to the tissues, plasma lactate concentrations begin to increase as a product of anaerobic metabolism.⁸ The value of lactate measurement is dependent on its ability to indirectly measure the severity of anaerobic metabolism in various states of shock. Because of the emergent state of shock patients and the rapid fluctuations in hemodynamic status during treatment, the clinical usefulness of lactate measurement is dependent on the ability to acquire accurate results rapidly. Rapid lactate measurement would allow clinicians to evaluate trends and institute appropriate therapy in a timely fashion once inadequate tissue oxygenation was identified. Due to the added value of serial evaluation of plasma lactate concentrations, cost may also play a major factor in the clinical utility of lactate measurements.

One portable lactate analyzer^a (device A) has previously been shown to be accurate, reliable, and cost-effective when used in human and equine intensive care patients.^9–13 The second analyzer^b (device B) has been used in dogs for point-of-care (POC) measurements of blood gases and lactate; however, an entire blood gas measurement must be obtained to evaluate lactate values, making serial measurements more expensive. The goal of this study was to compare two portable blood lactate analyzers and a reference laboratory (REF) to determine if these bedside monitors are reliable in the clinical setting.

Materials and Methods

Three lactate analyzers were used in this study. Device A measures lactate via reflectance photometry of a colorimetric lactate-oxidase mediator reaction of the plasma portion of whole blood. Device B measures lactate amperometrically, and the lactate is first converted to pyruvate and hydrogen peroxide. The hydrogen peroxide is further oxidized at a platinum electrode, producing a current proportional to the sample lactate concentration. The REF^c has a methodology similar to device B. The results from device A were recorded using both the “plasma” and “blood” settings (determined by proprietary algorithm). Both portable analyzers were calibrated daily according to the manufacturer’s instructions, and the manufacturer's recommendations for storage and handling of the samples to be analyzed were strictly applied. The REF was operated at the Comparative Neuromuscular Laboratory at the University of California-San Diego, CA according to standard operating protocol.

Samples were collected between Jan 1, 2005 and Jun 2006. All samples were collected and all POC testing was performed by one of two investigators (either M.K. or B.G.). Samples were obtained from a wide range of canine patients in the small animal intensive care unit at the University of Missouri Veterinary Medical Teaching Hospital. Samples were collected from hospitalized patients based on an attending clinician’s request in a clinical setting with owner permission obtained for all blood draws, diagnostics, and treatments. The samples were collected from both hemodynamically stable and unstable dogs to ensure that a variety of lactate concentrations were measured to evaluate performance of the instrument at both high and low lactate concentrations. All blood samples were venous, and all were obtained either via venipuncture or through a central catheter. Each sample was obtained with only a single blood draw to account for different levels of stress, restraint, or patient shivering/trembling that may lead to variance in lactate levels in samples from consecutive blood draws in the same patient. All 125 samples had plasma lactate concentrations measured on whole blood on the Accutrend lactate analyzer. Plasma lactate was measured on device B on all samples, but one measurement was below the measurement capability of device B (< 0.3 mmol/L) and that sample was not used in the final statistical analysis. A subset of the plasma samples were sent to the REF for comparison. The packed cell volume (PCV) was also determined on all samples. Five samples of varying lactate values (collected from apparently stable and unstable patients) were specifically chosen to have whole blood analyzed with device B 5–10 times each, depending on sample quantity, to evaluate the precision and intra-assay variation of the instrument. Due to increases in lactate that can occur from in vitro lactate production by erythrocytes, those five samples were run in immediate succession to decrease the amount of time blood was stored between sampling.³

A 3 mL sample of venous blood was collected from each patient then immediately placed in lithium heparin tubes^d. One drop of whole blood from the heparinized tube was assayed using device A within 10 min of collection. A standard pipette was held completely vertical in an attempt to standardize drop size when using device A. The heparinized tube of blood was then centrifuged to separate the plasma. The plasma was then assayed using device B. For the subset of samples collected for the REF, an additional 2 mL of blood from the same blood draw was placed in sodium fluoride/potassium oxalate tubes^e as instructed by the REF to halt red blood cell glycolysis. Those samples were immediately centrifuged and the plasma was separated and frozen (between −18°C and −20°C). The frozen samples were then sent to the REF by an overnight service on frozen gel packs to be assayed. Samples were not obtained from patients currently taking medications containing bromide or from patients that recently received a hemoglobin-based oxygen-carrying fluid^f because those products are known to interfere with plasma lactate measurement according to the manufacturer^a.

Statistical Analysis

The primary statistical method was the concordance correlation coefficient (CCC), a measure of the degree to which pairs of values fall on the 45° line (i.e., the line y = x).¹ The CCC measured the extent to which two methods (i.e., device A compared with REF and device B compared with REF) agreed in regards to identical results. In addition to giving a point estimate of the CCC, 95% confidence interval (95% CI) estimates were also established for device A and REF as well as for device B and REF. Scatter plots of the data with a 45° line included for reference were also generated. Relative differences between the two POC analyzers and the REF were determined. Clinically, a difference of 20% between lactate levels was considered to be a significant difference in evaluation of lactate. An indicator variable was defined (a 0–1 variable) that takes on the value 1 if the absolute relative difference exceeds 20% and takes on the value 0 otherwise. The McNemar test was used to compare the proportions of times the absolute relative difference exceeded 20%. To examine the effect of PCV on the measured values or differences (or relative differences) between device A and the REF, scatter plots of the data, as well as correlations between PCV and the other variables, were determined. Because some of the relative differences gave extreme values, Spearman rank correlations were used rather than Pearson correlations. A partial correlation coefficient between the difference and PCV, controlling for the variable REF, was also determined. The variability in repeated measures on device A was quantified by determining coefficient of variation (CV). All statistical analyses were performed using a commercial statistical software package^g.

Results

In total, 125 samples were collected and analyzed using the two portable blood lactate analyzers, and 86 samples were sent to the REF. One of the 86 samples sent to the REF, one was not included in statistical analysis because the lactate level was below the measurement capability of device B (< 0.3 mmol/L). Samples were collected from 49 different breeds. The most common breed was the Labrador retriever (18.4% of the population), followed by mixed-breed dogs (16.8% of the population). The ages of the dogs ranged from 2 mo to 16 yr (mean, 6.5 yr; median, 7 yr). The body weights ranged from 1.3 kg to 77.1 kg (mean, 27.6 kg; median, 25.9 kg).

The CCC for device A (on the plasma setting) and the REF was 0.949 (95% CI, 0.923–0.966). The CCC of the same samples when recorded on the whole blood setting for device A was lower (0.928; 95% CI, 0.901–0.948). For all additional statistical analyses, the value obtained using the plasma setting on device A were used. For device B and the REF, the CCC was 0.990 (95% CI, 0.985–0.993). Device B had better agreement with the REF than device A did (Figures 1, 2). When looking at the graph of device B versus device A (Figure 3), the values from device A tended to be higher, and the two analyzers were not in good agreement with each other. Regarding the relative differences (between either of the POC analyzers and the REF) exceeding 20%, 8 of 85 samples (9.4%) exceeded a relative difference of 20% for device B and 48 of 85 samples (56.5%) exceeded a relative difference of 20% for device A (Figures 4, 5). The difference in those proportions was statistically significant (P < 0.0001).

View Figure

• Download as Powerpoint
• Download Figure

View Figure

• Download as Powerpoint
• Download Figure

View Figure

• Download as Powerpoint
• Download Figure

View Figure

• Download as Powerpoint
• Download Figure

View Figure

• Download as Powerpoint
• Download Figure

PCVs ranged from 9% to 66% (mean, 38%). For the relationship between the difference of device A from the REF and the PCV, correlations were not significant.

Finally, precision data for device A have been provided in Table 1. The CV for each sample resulted in values ranging from 1.9 to 10.3 (mean, 5.6).

Table 1 Repeated Lactate Measurement of the Same Plasma Sample Collected from Five Dogs with Either Normal or Elevated Plasma Lactate Concentrations using Device A

*

Plasma lactate was measured at various time points (min postcollection) as indicated.

†

These samples were run on the same machine repeatedly in rapid succession, around one minute apart (length of time to run sample). The testing was started within 5 minutes of drawing sample.

§

The coefficient of variation for each sample was evaluated to examine the variability in repeated assays of the blood from the same sample taken under clinic conditions. The mean coefficient of variation was 5.6.

CV, coefficient of variation.

View Fullscreen

Discussion

Measurement of lactate levels can be a valuable indirect measurement of anaerobic metabolism, and lactate levels are routinely used as a determinant of hypoperfusion, either local or systemic. Anaerobic conditions can change quickly either as shock states progress or treatment is administered, making it important clinically to be able to measure lactate as a cage-side POC test to rapidly identify elevated anaerobic metabolism in real time. In this study, device A proved to be easy to operate, and results were displayed 1 min after application of the sample.

In human patients, lactate concentrations correlate to the degree of systemic hypoperfusion in shock states.² Traditionally, in human medicine, mild shock states lead to lactate concentrations between 2.5 mmol/L and 5 mmol/L (reference range, < 2.5 mmol/L). Moderate shock states result in lactate concentrations between 5 mmol/L and 10 mmol/L, and in severe decompensated shock, lactate concentrations can exceed 10 mmol/L.¹⁴ In this study, 71 of 125 samples measured using device A had lactate levels < 2.5 mmol/L, 37 of 125 samples were between 2.5 mmol/L and 5 mmol/L, 13 of 125 samples were between 5 mmol/L and 10 mmol/L, and 4 of 125 samples were > 10 mmol/L. There was a disproportionate distribution of lactate values because the majority of lactate values were < 2.5 mmol/L. Although included patients had a variety of different stages of cardiovascular compromise (determine based on physical exam findings and blood pressure measurements), their lactate concentrations were often either normal or close to normal. This may indicate a potential lack of sensitivity of global lactate measurement in determining degrees of shock states.

With treatment, lactate concentrations should drop as tissue perfusion improves. In this way, serial lactate monitoring may be used to guide resuscitation efforts. In humans, various studies have demonstrated that failure to lower lactate concentrations after resuscitative efforts is associated with a poor prognosis, and that prognosis improves in situations where lactate concentrations significantly decrease after treatment.³ A POC test that is simple, economical, and precise enough to make repeated bedside measurements would be a very useful tool in the intensive care unit. The variability for device A (mean CV was 5.6) can possibly be explained by considering that the samples were repeated by using a “drop” of blood, which was done by using a standard pipette held vertically to attempt to standardize the size of the drop. The size of the drop could have varied slightly, but it did replicate how a sample would be run in a clinical setting. Further attempts to standardize the exact amount of blood in the drop were not performed. It is possible that the size of the drop could affect the results of device A, which may account for the variability in the results. A previous study evaluating device A in human obstetrics revealed that drop size may affect lactate values, especially if a sample volume < 20 μL was used.⁹ Another study evaluating the use of device A in dogs compared the CV when one drop was used to using fixed 25 μL samples. The CV for the exact measurement in that study was 5.3, whereas the CV when a drop was used ranged from 14% to 15%, depending on the lactate value.¹⁵ The CV for the current study (5.6) was comparable to the CV of the fixed volume samples in the previously cited study. This may be explained by the fact that even though a precise amount of blood was not measured in the current study, the technique and size of the pipette used to add one drop was consistent, so the drop size may not have been as variable as in the previous study. However, it is unknown how the variable drop samples were created in the previous study. Also, although the samples were run in immediate succession in the study described herein, lactate generation by erythrocytes in the sample could theoretically account for some variation in lactate measurements over time. When evaluating Table 1, it is evident that the variability would not likely result in any clinically relevant differences; however, care must be taken in interpreting small changes in lactate measurements when trending lactate values in a patient and evaluating response to treatment.

In this study, both portable blood lactate analyzers demonstrated good agreement with the reference instrument, although device B was in much better agreement. Caution must be used when comparing the two POC instruments against the REF because the REF has not been established as the gold standard in the dog. Device B and the REF would be predicted to have a better agreement due to the similarity of their methodology (i.e., an amperometric technique using a platinum electrode) as opposed to the reflectance colorimetric methodology of device A. A recent study evaluating four POC meters in dogs also demonstrated agreement of both POC instruments with a REF^h. That study found better agreement with device B and other amperometric meters than device A, the latter of which was the only meter to use reflectance photometry in that study.¹⁶ The difference in methodologies may explain the increased differences between device A results and the REF compared with device B and the REF. The results generated by device A were higher than both device B and the REF, and device B results were slightly lower (on average) than the REF. Because there was more than 20% relative difference in device A versus the REF in > 50% of the samples tested, it is not advisable to use the two instruments interchangeably in the clinical setting. Also, while differences in values between the REF and device B were unlikely to lead to clinically significant differences in diagnosis or treating patients, there were a few points of considerable disparity when comparing device A with the other two machines (Figures 1–3). For example, one sample measured 5.84 mmol/L and 5.98 mmol/L (considered consistent with a moderate state of shock) on device B and the REF, respectively, but only 2.3 mmol/L on device A (which is considered within normal limits). Another sample measured within normal limits on device B and the REF (1.5 mmol/L and 1.69 mmol/L, respectively) but measured 3.8 mmol/L on device A.

Device A uses similar technology to many blood glucose units in that there are multiple layers through which the sample must diffuse. Although the original sample for device A is whole blood, the red blood cells are separated prior to reading the sample. Thus, evaluating a sample with a significantly elevated PCV may result in falsely low readings due to a decreased amount of plasma available to be analyzed. The effect of elevated PCV on lactate measurement with device A has been well documented in previous studies.⁹ In a previous study evaluating device A in horses, an elevated PCV (> 53%) caused values to be falsely decreased.¹¹ In that study, there was no obvious relationship between lactate measurements and PCVs. This may have been due to an insufficient number of patients with high PCVs to expose such a trend. In the current study, only 10 of 125 samples had PCVs > 53%.

Limitations of this study include the sample type and the size of the sample used on device A. According to manufacturer’s recommendations, whole blood was used on device A. Although the test strip filters the red blood cells, allowing for measurement of plasma, the effect of using plasma on the test strip was not evaluated. The use of actual plasma as a sample has been shown to be more accurate over a wider range of lactate and PCV values in horses.¹² Using plasma as a sample for device A should not be confused with using the plasma setting, which was used in this study as previously mentioned. Also, an exact measurement of a fixed volume of blood was not attempted; therefore, the impact of drop size on precision could not be determined. Finally, the order in which the samples were run may be a source of bias because device A was consistently assayed first before centrifuging the samples to separate plasma and assaying the samples on device B and freezing the samples to be assayed on the REF.

Conclusion

Although there is agreement between the three methodologies as measured by the CCC, there are significant differences between device A’s readings and those of device B and the REF. The three machines should not be used interchangeably. Also, although trends in lactate measurements can add valuable information as to how a patient is responding to treatment, small changes in lactate measurements when using device A to evaluate response to treatment or trends over time should be interpreted with caution.