Department of Medicine, Faculty of Veterinary Medicine, Khon Kaen University, Khon Kaen, 40002 Thailand; Department of Veterinary Clinical Medicine, College of Veterinary Medicine, University of Illinois at Urbana-Champaign, Urbana, Illinois, 61802, USA
Reasons for performing study: Some methods of lactate (LA) measurement have not been validated appropriately for use in horses.
Objectives: To validate 2 LA analysers (YSI 2300 Stat Plus and TDx Lactic Acid Assay) for use with equine plasma and to compare plasma [LA] determined by the 2 methods.
Methods: Both instruments were evaluated for linearity, parallelism, recovery and precision using serial dilutions of standard LA solutions and equine plasma and then comparing results with linear regression or paired t tests. Plasma [LA] results were compared in 275 blood samples collected from horses exercising at various intensities using Bland-Altman analysis. Level of significance was P<0.05.
YSI exhibited good linearity for both LA standards and equine plasma (P<0.05) at 0–30 mmol/l. TDx had good linearity at 0–12 mmol/l (P<0.05); with LA standard solutions >12 mmol/l and with equine plasma, linearity was decreased. YSI exhibited good parallelism between LA standards and equine plasma LA measurements throughout the 0–30 mmol/l range (P>0.05). Parallelism was poor with TDx (P<0.05). Mean ± s.d. % recovery was 101.7 ± 3.4% for YSI (acceptable) and 110.6 ± 8.4% for TDx (unacceptable). Within-run and mean between-run coefficients of variation (CV) for plasma samples tested from 3.3–29.5 mmol/l were 0.4–3.0% for YSI. CVs for samples tested from 2.8–8.0 mmol/l were 17.4–24.1% for TDx. In 275 plasma samples, [LA] ranged from 0.1–42.7 mmol/l and 0.3–50.6 mmol/l for the YSI and TDx methods, respectively. The difference in plasma [LA] determined by the 2 methods was −1.0 ± 3.2 mmol/l, documenting that the TDx overestimated the YSI results by a mean value of 1 mmol/l.
Conclusions: It was concluded that the YSI method was a reliable method for measuring equine plasma [LA] from 0–30 mmol/l. The TDx method was found not to be suitable for use with equine plasma due to greater variability in measurements (high CV).
Many experiments have used either blood or plasma lactate in determinations of fitness, training and racing performance in different intensities of exercise in horses (Persson 1983; Harris and Snow 1992; Rasanen et al. 1995; Desmecht et al. 1996; Jahn et al. 1996). In addition, some studies have demonstrated the relationship between a horse's fitness or exercise intensity and the horse's plasma biochemistry (including lactate) during exercise (Casini and Greppi 1996; Guhl et al. 1996; Jahn et al. 1996; Werkmann, et al. 1996). They presented useful information concerning the use of lactate in equine exercise physiology. The ease of sampling, in addition to the availability of automatic lactate analysers and test kits, has contributed to the increasing popularity of measuring lactate as a routine parameter to be considered for both clinical measuring and examining exercise physiology. Most lactate analysers use the concept of enzymatic analysis (method for automatic analysers) as described by Bergmeyer (1974) in measuring blood or plasma [LA]. Plasma can be measured spectrophotometrically or fluorimetrically using the lactate dehydrogenase (LDH) catalysed reaction with nicotinamide adenine dinucleotide (NADH) (Lactate + NAD+↔LDH Pyruvate + NADH + H+). In both methods, the decrease in extinction due to oxidation of NADH in the reaction with pyruvate as substrate is measured (Bergmeyer 1974).
The YSI 2300 STAT PLUS1 has been widely used in equine exercise physiology research (Eaton et al. 1995; Foreman et al. 1996; Schott et al. 1997; Hyyppä 2007). The YSI Lactate Analyzer is an automated lactate analyser that uses an enzymatic method and sensor technology to measure plasma lactate concentration. The sensor probe consists of an electrode that is fitted with a 3 layer membrane containing immobilised oxidase in the middle layer and a stainless steel electrode (platinum anode). The face of the probe, covered by the membrane, is situated in a buffer-filled sample chamber into which a sample is injected. When L-lactate diffuses through the membrane and contacts the immobilised oxidase enzyme, it is rapidly oxidised to pyruvate and hydrogen peroxide (H2O2), which is then oxidised at the platinum anode, producing electrons. A dynamic equilibrium is achieved when the rate of H2O2 production and the rate at which H2O2 leaves the immobilised enzyme layer are constant and is indicated by a steady state response. The electron flow is linearly proportional to the steady state H2O2 concentration and, therefore, to the concentration of lactate.
The TDx Lactic Acid assay2 has been used in human exercise physiology research (Cox 1994; Steppich et al. 2000; De Palo et al. 2001). The TDx assay is a reagent system for the quantitative measurement of plasma lactate concentration that utilises radiative energy attenuation (REA) technology (Shaffar and Stroupe 1983; Refer to the TDxFLx System Operation manual). In the assay, a chromophore is generated by the coupled enzymatic reactions of lactate dehydrogenase (LDH) and diaphorase. The amount of chromophore produced is directly proportional to the lactic acid concentration of the sample.
The aim of this study was designed to investigate the suitability of these 2 lactate analysers within the ranges of linearity stated by the manufacturers (0–30 mmol/l and 0–12 mmol/l for YSI and TDx, respectively) for use with equine plasma and to compare plasma lactate concentrations determined by the 2 methods. In general, a high range of lactate concentrations is demanded for the measurement in equine exercise physiology and accuracy within that range is required for valid equine research. We chose the TDx lactic acid assay to compare with the YSI because it had a narrow range of lactate concentration and it was in common use in our College for other experimentation and, if accurate for use in horses, it was felt that a single assay might be adopted for lactate measurement in some of our College's laboratories.
Materials and methods
Linearity: Lactate standards and equine plasma collected in sodium fluoride/potassium oxalate tubes were diluted with buffer provided by the manufacturer to obtain different lactate concentrations within the range of linearity stated by the manufacturers (0–30 mmol/l and 0–12 mmol/l for YSI and TDx, respectively). The measured values were then plotted against the actual values.
Parallelism: The equine plasma samples were diluted successively with buffer from each manufacturer to provide different lactate concentrations covering the range of linearity for each method. In each method, selected equine plasma had different lactate concentrations to the reported upper limit for lactate measurement. For the YSI method, equine plasma was used and diluted successively with one part of YSI calibrator (5 mmol/l lactate concentration) to achieve different lactate concentrations (20.2, 12.6, 8.8, 6.9, 6.0 and 5.5 mmol/l) within the desired range of linearity. For TDx assay, equine plasma was used and diluted successively with one part of TDx calibrator (0 mmol/l lactate concentration) to achieve different lactate concentrations (9.5, 4.8, 2.4, 1.2 and 0.6 mmol/l) that cover the range of linearity. The measured values were then plotted against the actual values and the parallelism was measured as compared to the standard linearity line.
Recovery: This test was performed by using 3 plasma samples separated from venous blood collected into sodium fluoride/potassium oxalate tubes from 3 healthy horses at rest to obtain baseline samples with low concentration of lactate. These samples were spiked with known amounts of an analyte, a pure standard of stock lactate at high concentration to achieve plasma lactate in different concentrations. For the YSI method, lactate standard at concentrations of 30 mmol/l was added to equine plasma samples to obtain plasma lactate at 1.25, 2.5, 7.5 and 15.0 mmol/l. For the TDx, lactate standard of 12 mmol/l was added to equine plasma samples to obtain plasma lactate at 0.75, 1.5, 3 and 6.0 mmol/l. The formula for calculating concentration of standard added, concentration recovered and percentage recovery was used according to Bakes-Martin (1993) as follows:
Lactate measurements were obtained in triplicate before and after spiking and the percentage recovery was calculated. Recovery of 95–105% was considered acceptable.
Precision: Precision measurements were made to test the repeatability and reproducibility of both instruments. Initial validation of a method should give an indication of the within-run (repeatability) and between-run (reproducibility) variation. Within-run variation was determined by performing 10 replicates of equine plasma samples at different concentrations covering the range of the linearity line (2.5, 5.5, and 9.5 mmol/l for TDx and 2.5, 5.5, 16.1, 20.5 and 28.1 mmol/l for YSI). According to Bakes-Martin (1993), precision is expressed as the coefficient of variation (CV): CV = (s.d./mean) × 100. The usual criterion for acceptance in repeated laboratory measurement for CV should be <5%. For each method, 6 replicates were studied for each concentration. The within-run variations were determined by calculating the mean, s.d. and CV at each lactate level. Between-run variation was determined with equine plasma samples at different concentrations for 3 different runs (3 days). The mean, s.d. and CV were calculated and the average between-run variation determined.
Comparison of plasma lactate concentrations determined by 2 methods
The study consisted of 275 blood samples from healthy Standardbreds, Thoroughbreds and Quarter Horses performing at different exercise intensities. A 10 ml sample of blood was collected from each horse into evacuated tubes containing sodium fluoride and potassium oxalate. Samples were centrifuged at 3000 g for 5 min, the plasma aspirated, placed into plain tubes with sealed caps and kept frozen (at −20°C) before analysis. Samples were thawed rapidly at 25°C and analysed promptly and randomly and plasma lactate concentrations were measured following instructions from each manufacturer. Calibration of the lactate analysers was performed by using a standard solution supplied by each manufacturer. Samples in which the plasma lactate concentration was greater than the upper limit of measurement were diluted with the buffer supplied by the manufacturer. The plasma lactate concentrations measured by YSI and TDx methods were then plotted as a scatter plot.
Linear regression analysis was used in the validation section to describe the linearity and parallelism tests of lactate standard and equine plasma lactate measured with 2 methods. Linear regression was also used to test the difference between the linearity lines and the line of identity, and the difference between the slopes of lactate standard curves and equine plasma lactate curves for the parallelism tests at the 95% confidence interval. The association between the plasma lactate concentrations measured with the 2 instruments was graphically expressed in scatter plots. Additionally, for comparison of the YSI and TDx methods, the lactate concentration difference and the mean lactate concentration in each sample measured with the 2 methods were evaluated by the graphical method of Bland and Altman (1986). The mean, s.d. and limits of agreement (confidence interval) of the difference in plasma lactate concentration were calculated. The confidence interval was defined as the mean ± s.d. difference, which represented 95% of the difference.
The linearity data are shown in Figure 1. The YSI method exhibited good linearity for both lactate standard and equine plasma samples throughout the manufacturer's stated range of 0–30 mmol/l (Fig 1a, b). The linear regression of the data was as follows: measured [Plasma lactate] ([pLA]m) = 1.01 actual [Plasma lactate] ([pLA]a) − 0.26, and [pLA]m = 1.00 [pLA]a + 0.43 (r = 0.999, P<0.05) for the lactate standard and equine plasma lactate, respectively. For the TDx method, good linearity was found only within the range stated by the manufacturer (0–12 mmol/l) for the lactate standard: [pLA]m = 0.99 [pLA]a + 0.18 (r = 0.999, P<0.05, Fig 1c). At lactate standard concentrations >12 mmol/l (diluted with buffer), the linearity was decreased: [pLA]m = 0.82 [pLA]a + 0.36 (r = 0.999, P<0.05). When used with equine plasma, linearity was decreased (Fig 1d): [pLA]m = 0.65 [plasma lactate]actual+ 0.72 (r = 0.999, P<0.05), but there was no significant difference from the line of identity (P>0.05).
Parallelism results are shown in Figure 2. The YSI method exhibited good parallelism throughout the manufacturer's stated range of 0–30 mmol/l (Fig 2a). The linear regression of the data was [pLA]m = 1.01 [pLA]a − 0.26 (r = 0.999, P<0.05) for the lactate standard, and [pLA]m = 1.00 [pLA]a − 0.26 (r = 0.999, P<0.05) for equine plasma. There was no significant difference between slopes of lactate standard curve and equine plasma lactate curve measured by the YSI method (P>0.05). Parallelism was not found with the TDx method (Fig 2b). The regression lines were [pLA]m = 0.99 [pLA]a + 0.18 (r = 0.999, P<0.05) and [pLA]m = 0.75 [pLA]a + 0.20 (r = 0.999, P < 0.05) for the lactate standard and equine plasma, respectively, but there was significant difference between the slopes of the lactate standard curve and equine plasma lactate curve measured with the TDx method (P<0.05).
The recovery data for the 2 methods are shown in Table 1. The YSI method exhibited good recovery. The mean lactate recovery was 94.8% at 1.25 mmol/l, 104.5% at 2.5 mmol/l, 104.3% at 7.5 mmol/l and 103.2% at 15 mmol/l. The mean ± s.d. percent recovery was 101.7 ± 3.4%. For the TDx method, the percent recovery was not in the range of acceptance (95–105%). The mean lactate recovery was 92.0% at 0.75 mmol/l, 119.9% at 1.5 mmol/l, 120.0% at 3.0 mmol/l and 110.6% at 6.0 mmol/l. The recovery was 110.6 ± 8.4%.
Table 1. Recovery data for the YSI and TDx methods
Lactate concentration (mmol/l)
Mean = 101.7
Mean = 3.4
Mean = 110.6
Mean = 8.4
The mean plasma lactate concentrations (mmol/l), within- and between-run CV are shown in Table 2. For the YSI, the within-run CV ranged from 0.4–3.0% at mean concentrations of 3.3–29.5 mmol/l. The average between-run CV from 3 days ranged from 0.4–3.0% at mean concentrations of 3.3–29.5 mmol/l. For the TDx, the within-run CV ranged from 17.4–24.1% at mean concentrations of 2.8–8.0 mmol/l. The average between-run CV from 3 days ranged from 17.9–24.0% at mean concentrations of 2.8–8.0 mmol/l.
Table 2. Mean plasma lactate concentrations (mmol/l), within-run, and between-run coefficient of variation (CV) measured with YSI and TDx methods
Mean lactate concentration (mmol/l)
Plasma lactate samples (n = 275) obtained at rest, during exercise (via pulmonary artery catheters), or at 5 min post exercise were measured by the YSI and TDx methods. Lactate concentration ranged from 0.1–42.7 and 0.3–50.6 mmol/l for the YSI and and TDx methods, respectively. Scatter plot between the YSI and TDx plasma lactate concentrations is shown in Figure 3. Comparison of YSI and TDx plasma lactate concentrations was used following Bland and Altman graphical method (1986). The difference in plasma lactate concentrations between the 2 methods was mean ± s.d., −1.0 ± 3.2 mmol/l (Table 3). The TDx method overestimated the YSI method in measuring plasma lactate by a mean value of 1 mmol/l. The limit of agreement was approximately 6.3 mmol/l above and below the mean of difference (Fig 4).
Bioanalytical methods must be validated prior to and during use to have confidence in the results (Buick et al. 1990; Causey et al. 1990; Bakes-Martin 1993). In this study, we found that the YSI had a very good linearity, parallelism, recovery and precision when compared to the TDx method. When these methods were used with equine plasma samples, linearity was found only with the YSI method. For the TDx, there was considerable variability in linearity when the lactate standard concentrations were >12.0 mmol/l and there was considerable variability in linearity when used with equine plasma samples even within the manufacturer's claimed range of 0–12 mmol/l (Fig 1). Parallelism was found only with the YSI method. The recovery of 95–105% was considered acceptable for this experiment (Bakes-Martin 1993). The recoveries of the YSI method were acceptable (mean ± s.d. = 101.7% ± 3.4), but the TDx was unacceptable (mean ± s.d. was 110.6% ± 8.4). The YSI method showed very good repeatability and reproducibility when used with equine plasma samples. These results are similar to those of Evans and Golland (1996), but ours were performed at higher plasma lactate concentrations.
The different results between the YSI and TDx in this study may come from the inherent methodological variations and the variability of the TDx when used with equine plasma may be compounded by the use of serial dilutions until a value is obtained in the 0–12 range claimed by the manufacturer. Those serial dilutions may add potential error to the analysis. Basically, these 2 instruments were originally designed for use with human biological samples, but there is a different matrix in blood among species (Buick et al. 1990).
In comparing the YSI and TDx methods, the usefulness of application of these 2 methods was determined by how closely one method predicts another and precision of the prediction (limits of agreement or confidence interval) (Bland and Altman 1986). The observed bias between 2 methods could be useful if the limits of agreement are narrow indicating that one technique is reliably different in magnitude than the other. In this study, a confidence interval of 95% (2 s.d.) of the difference was used. The plasma lactate concentration measured by the TDx method overestimated the YSI method. The bias was associated with a confidence interval of ± 6.3 mmol/l (± 2 s.d.). For example, given plasma lactate concentration of 20 mmol/l measured by the TDx method, there is a 95% chance that the plasma lactate concentration measured by the YSI method would be between 13.7 and 26.3 mmol/l. Therefore, the judgement regarding plasma lactate concentration measured by the TDx would be erroneous and lead to misinterpreted high lactate concentration.
The YSI assay can be used for measurement of equine plasma lactate concentrations throughout the manufacturer's stated range of 0–30 mmol/l, which is a suitable test for measuring lactate in resting horses, exercise studies and evaluation of diseased horses. At rest, equine blood lactate concentrations are normally <1 mmol/l (Hodgson 1996) and, in this study, the YSI showed a greater variability in measuring lactate concentration at low concentrations. Elevation of lactate concentrations >1 mmol/l have been observed in sedentary diseased horses, i.e. horses with laminitis (Garner et al. 1978), muscular degeneration (Moore et al. 1976) and after prolonged recumbency due to anaesthesia (Linsay et al. 1980). In exercising horses, lactate production is closely related to the intensity of the exercise (Desmecht et al. 1996) and lactate concentration continues to increase to a peak concentration around the end of exercise up to >40 mmol/l (Marlin et al. 1991). We verified that the YSI can be used to measure accurately plasma lactate concentrations up to 30 mmol/l. At concentrations >30 mmol/l, we recommend that the plasma samples be diluted with buffer before analysis. It was concluded that the YSI method was a reliable method for measuring equine plasma [LA] from 0–30 mmol/l. In this study, we tested the TDX analyser beyond its published range, because many equine plasma samples have LA concentrations >12 mmol/l and accuracy in that range is, therefore, required for equine research. We verified that the analyser was not accurate in that range, rendering it not a valuable tool for many equine plasma samples unless they are screened, found to be above 12 and then serially diluted until a value is obtained in the 0–12 range. Those dilutions will add additional work and potential error to the analysis and we confirmed that this method was indeed not accurate outside the manufacturer's stated range. The TDx method had a greater variability in measurements (high CV) even within the manufacturer's stated range of validity.
Conflicts of interest
The authors have declared no potential conflicts.
1 Yellow Springs Instruments, Yellow Springs, Ohio, USA.
2 Abbott Laboratories, Abbott Park, Illinois, USA.