Veterinary Clinical Pathology ISSN 0275-6382
ORIGINAL RESEARCH
Validation of the handheld Lactate-Pro analyzer for measurement of blood L-lactate concentration in cattle bastien Buczinski, Elizabeth Dore , Guillaume Boulay, David Francoz Se al, Montre al, QC, Canada Department of Clinical Sciences, Faculty of Veterinary Medicine, University of Montre
Key Words Biochemistry, bovine, dairy, handheld analyzer, lactatemia Correspondence S. Buczinski, Faculty of Veterinary Medicine, University of Montreal, CP5000, SaintHyacinthe, QC Montreal J2S 7C6, Canada E-mail: [email protected] DOI:10.1111/vcp.12185
Background: Blood L-lactate concentration (LAC) can be used for various diagnostic purposes in cattle. As multiple handheld analyzers for LAC exist, it is important to validate their use in cattle in comparison with reference laboratory blood analyzers. Objectives: The objectives of this study were to validate the handheld Lactate Pro meter (LacP) including reproducibility, and compare the measurements with the StatProfile (StatP) as a gold standard. In addition, diagnostic sensitivity and specificity, and the impact of HCT on LAC measured by both analyzers were assessed. Methods: A cohort of 64 cattle with acute medical and surgical conditions was studied. Whole blood samples in heparin lithium tubes were analyzed upon arrival with both StatP and LacP. Twenty-three samples were immediately retested to assess intra-assay coefficient of variation (CV). The HCT values were also recorded. Results: The LAC using LacP was highly correlated with the StatP (r = 0.9736 [95% confidence interval [CI]: 0.9562–0.9841]). The LacP underestimated LAC (mean difference:0.9 mmol/L, 95% CI:3.1 mmol/ L to 1.3 mmol/L). The intra-assay CV was excellent (4.77%). No significant correlation was observed between LacP or StatP and HCT (P = .39 and .09, respectively). Sensitivity and specificity for LacP were 91.7% (95% CI: 76.4–97.8%) and 100% (83.4–100%, cutoff of 4 mmol/L), and 78.6% (58.5–90.9%) and 100% (87.0–100%, cutoff of 6 mmol/L). Conclusions: The LacP handheld lactate meter can be used safely and reliably cow-side, although it underestimates LAC value when compared with a standard laboratory analyzer especially for LAC ≥ 10.0 mmol/L. The LAC value was not influenced by HCT in this study.
Introduction L-lactate is a sensitive marker of tissue perfusion as it is mainly produced during anaerobic glycolysis. Blood lactate may increase in various conditions including shock secondary to hypovolemia, endotoxemia, or heart failure, and is frequently used in diagnostic workups in human and veterinary medicine.1 By contrast to L-lactate which is produced by the mammalian cell, D-lactate, the other enantiomeric form of lactate, is produced by bacterial fermentation in the gastrointestinal tract of cattle.2 D-lactic acidosis has been implicated in various syndromes in adult and neonate cattle.2,3 All handheld analyzers of lactate assess L-lactate. A
rapid point-of-care analysis system for D-lactate assessment is currently not available. In cattle, rising blood L-lactate concentration (LAC) has been correlated with an unfavorable prognosis in cases of abomasal volvulus or right displaced abomasum4,5, or bovine respiratory disease.6 Specifically, in beef calves with bronchopneumonia6, LAC > 4 mmol/L was a reliable prognostic marker for animal death within 24 hours. We recently found that LAC > 6 mmol/L was associated with higher risk of poor outcome in the 30 days postsurgery in dairy cows with right abomasal dilation/volvulus.5 LAC has also been used for assessing the severity of neonatal calf distress shortly after birth.7 In horses, several studies have shown that LAC in whole blood can be underestimated
Vet Clin Pathol 43/4 (2014) 567–572 ©2014 American Society for Veterinary Clinical Pathology
567
Validation of handheld lactate meter in cattle
Buczinski et al
when measured using handheld analyzers in the presence of increased HCT.8,9 To date, scarce information is available concerning the correlation between HCT and LAC in cattle. One of the most important advantages of the handheld analyzers is that they can quickly assess LAC in a farm setting allowing more rapid and adequate patient management. Different handheld blood L-lactate analyzers are available for the clinician.4,6,10–12 Validation of these portative meters, including the determination of bias in comparison with standard chemistry analyzers, is of crucial importance particularly if LAC is used for prognostic assessments. The LactatePro (LacP) is a handheld L-lactate analyzer that gives a rapid result of LAC in the range from 0.8 to 23.3 mmol/L with only one drop of blood, and can be used easily at the farm. To date, its performance has been shown to be comparable to standard biochemistry analyzers in dogs13, cats14, and horses.11 However, the LacP has not been validated for cattle blood so far. The main objective of this study was to validate this device to measure LAC in cattle. Our hypothesis was that LAC is accurately estimated in cattle by the LacP analyzer in comparison with the StatProfile Critical Care Xpress (StatP) analyzer as a gold standard. A second objective was to assess the impact of HCT on LAC measurement by both analyzers.
Materials and Methods The study protocol was accepted by the ethical committee of the Faculty of veterinary medicine, University of Montr eal. In this prospective study, cows and calves admitted to the veterinary teaching hospital with acute medical and surgical conditions, including gastrointestinal diseases, various degrees of shock, or severe infectious processes, were eligible for the study. This study population was chosen (1) because of its anticipated variation of LAC values secondary to various degrees of hypoperfusion, shock, and infection, and (2) because the initial diagnostic workup of these animals routinely includes measurement of blood gas, pH, electrolytes, and L-lactate using a heparinized blood sample collected upon arrival. Gold standard measurements of LAC were performed using the StatProfile Critical Care Xpress (Nova Biomedical Corporation, Waltham, MA, USA). This analyzer uses an amperometric reaction to determine LAC. A whole blood sample of 130 lL is exposed to an electrode embedded with lactate oxydase. The lactate is converted to pyruvate and hydrogen peroxide. Hydrogen peroxide is oxidized
568
generating a current proportional to the LAC. This analyzer has an analytic range of 0.3–20.0 mmol/L. The data obtained were plotted in comparison with the values from the handheld device.13,15 The LactatePro (Arkray, Kyoto, Japan) is a portative handheld analyzer based on an amperometric method using an enzymatic reaction. The L-lactate reacts with the lactate oxydase, which reduces potassium ferricyanide to potassium ferrocyanide. The potassium ferrocyanide is then oxidized to potassium ferricyanide and produces an electric current proportional to lactate concentration of the sample. The result is displayed within 60 s after a drop of heparinized blood has been placed on the strip.
Estimate of random error for both methods To estimate the random error of analysis of both methods, a second analysis in both analyzers within 5 min allowed the assessment of the intra-assay coefficient of variation of both methods (CVLacP and CVStatP).
Clinical performance of the Lactate-Pro analyzer The efficiency of nonsurvival prediction by the LacP was determined for 2 different clinical cutoffs at 4 mmol/L for bronchopneumonia in calves6, and 6 mmol/L for right abomasal disorders in cows5 (determined by StatP analyzer as a gold standard). These 2 different cutoffs were chosen because of their clinical relevance in 2 major syndromes in bovine medicine. The clinical performance of the handheld analyzer was assessed evaluating sensitivity (Se) and specificity (Sp) of nonsurvivors detection at both cutoffs (4 or 6 mmol/L, respectively), and was designated SeLAC4, SpLAC4, SeLAC6, and SpLAC6. The formulae used were: (1) Sensitivity: SeLACCutoff (4 or 6 mmol/L, respectively) = number of samples with LAC using LacP ≥ 4 or 6 mmol/L, respectively/number of samples with LAC using StatP ≥ 4 or 6 mmol/L, respectively; (2) Specificity: SpLACCutoff (4 or 6 mmol/L, respectively) = number of samples with LAC using LacP < 4 or 6 mmol/L, respectively)/number of samples with LAC using StatP < 4 or 6 mmol/L, respectively. We defined acceptability criteria for the Se and Sp at a theoretical level of 85%. This arbitrary threshold was established based on mean values of different diagnostic tests used in human medicine.16 The 95% confidence intervals (CI) of these parameters were
Vet Clin Pathol 43/4 (2014) 567–572 ©2014 American Society for Veterinary Clinical Pathology
Buczinski et al
determined using a binomial distribution modeling as follows17: p 95% CI Se ¼ SeLAC4 or 6 1:96 ððSeLAC4 or 6 ð1 SeLAC4 or 6 Þ=nÞ; and p 95% CI Sp ¼ SpLAC4 or 6 1:96 ððSpLAC4 or 6 ð1 SpLAC4 or 6 Þ=nÞ; where SeLAC4 or 6 and SpLAC4 or 6 are the proportions found based on the contingency table and n, the total number of cases (n = 61 in this study). Due to the possible implication of this diagnostic test on individual management of patients, we calculated the negative and positive likelihood ratio (NLR and PLR, respectively). The NLR is defined as the number of true-positive patients (defined with the gold standard [StatP]) with a negative test result divided by the number of true-negative patients with a negative test result. The PLR represent the number of true-positive patients with a positive test result divided by the number of true-negative patients with a positive test result. One of the main interests of these parameters is that they do not depend on the prevalence of the outcome, but only on test Se and Sp, by contrast to predictive values.16 We aimed for a test that would have a good NLR (≤ 0.2)16 adequately identifying cows or calves with a favorable outcome (ie, survival). The NLR calculation was obtained from the formula: NLR = (1Se)/Sp as well as 95% CI.17 The PLR was also calculated using the formula PLR = Se/(1Sp) as an indicator of the odds ratio of nonsurvival based on the LacP analyzer classification. Where available, the HCT was determined by a hematology analyzer (Cell Dyn 3500, Abbott Laboratories, Mississauga, ON, USA) to assess its impact on the LAC values.
Statistical analysis All analyses were performed using commercial statistical software (SAS 9.2, Cary, NC, USA) including the graphics (MedCalc Software bvba, version 12.3.0, Mariakerke, Belgium). The relation between LAC determined by LacP and the StatP was determined using the Pearson correlation coefficient (r). A t-test was performed to determine if r was significantly different from 0. As comparison of different measurements using the Pearson correlation coefficient may have some limitations18, the agreement between the 2 methods was analyzed using the Bland–Altman analysis.19 The constant bias between the 2 methods and
Validation of handheld lactate meter in cattle
their limits of agreement (bias 2 SD) were also determined. The intra-assay coefficients of variation (CV) of both analyzers were calculated and expressed as a mean percentage. The combined inherent imprecision (CVI) was calculated using the formula for single measurements: p CVI ¼ ððCVLacP Þ2 þ ðCVStatP Þ2 Þ: The acceptability based on inherent imprecision of both tests was assessed using the mean value of both analytic methods and the difference between both methods, as well as both values corresponding to the 95% of the differences as previously recommended.20 As only 61 cases were used for this study, we expected that 95% of the differences between both methods would lie between (0 z 9 CVI 9 Mean of the 2 methods). The z value of 2.3 was used rather than the classical value of z = 1.96 for normally distributed data as previously recommended for data sets with < 70 samples.15 To assess the impact of HCT on the LAC measured by LacP and StatP analyzers, Pearson correlation coefficients were calculated and their statistical significance was determined as previously mentioned. The level of significance was set at P < .05.
Results A total of 64 cattle were eligible for inclusion in the study during the period from January 2011 to January 2012. The LAC range with the StatP analyzer was 0.4 mmol/L–> 20.0 mmol/L, and with the LacP analyzer < 0.8 mmol/L–19.9 mmol/L. Three animals were excluded from further analysis because they were out of range on one of the analyzers (2 animals with LAC < 0.8 mmol/L with LacP analyzer, and one with LAC > 20.0 mmol/L with StatP analyzer). Of the remaining 61 cattle, 45 had different acute gastrointestinal problems, 9 were downer-cows, one had bronchopneumonia, and 6 had no definitive diagnosis upon admission. The LAC using LacP analyzer ranged from 0.8 – 14.9 mmol/L (median: 4.9 mmol/L), and the LAC using StatP analyzer ranged from 0.6 mmol/L – 17.0 mmol/L (median: 5.7 mmol/L). The LAC from the LacP was highly correlated with LAC from the StatP (r = 0.9736, 95% CI: 0.9562– 0.9841; P < .01). The Bland–Altman analysis showed an excellent agreement between the reference method and the LacP when LAC was < 10 mmol/L (Figure 1). However, the LacP underestimated LAC compared
Vet Clin Pathol 43/4 (2014) 567–572 ©2014 American Society for Veterinary Clinical Pathology
569
Validation of handheld lactate meter in cattle
Buczinski et al
with the StatP analyzer (mean bias: 0.9 mmol/L, 95% CI from 3.1 mmol/L to 1.3 mmol/L). Differences between the 2 methods (LAC using LacP)-(LAC using StatP) ranged from 5.5 mmol/L to 0.4 mmol/L (median = 0.6 mmol/L). Twenty-three samples (from 0.9 mmol/ L – 12.6 mmol/L; median 3.6 mmol/L) were analyzed in duplicate with the LacP analyzer (Figure 2) and the StatP analyzer. The intra-assay CV for the LacP analyzer and the StatP analyzer was 4.77% and 2.69%, respectively. The inherent imprecision of both methods according to the formula (CVI = √((CVLacP)2 + (CVStatP)2) was 5.48%. Twenty-nine of 61 differences (48%) between the 2 tests were included in the inherent imprecision ranges (0 2.3 9 CVI 9 mean) (Figure 3). As less than 95% of the differences between both methods were within the imprecision ranges, this showed that the 2 methods were not identical and in fact reported different results for LAC in the same sample. For the 4 mmol/L cutoff, the LacP analyzer had a SeLAC4 of 91.7% (33 of 36 cases; 95% CI: 76.4–97.8%) and a SpLAC4 of 100% (25 of 25 cases; 95% CI: 83.4– 100%), and for the 6 mmol/L cutoff, SeLAC6 was 78.6% (22 of 28 cases; 95% CI: 58.5–90.9%) and SpLAC6 was 100% (33 of 33 cases; 95% CI: 87.0–100%) (Figure 4). The NLRLAC4 and NLRLAC6 were 0.08 (95% CI: 0.02–0.25) and 0.21 (95% CI: 0.11–0.44), respectively. As both SpLAC4 and SpLAC6 were 100%, it was not possible to calculate PLR.
Figure 2. Correlation of mean whole blood L-lactate concentrations (LAC) from 23 cattle measured in duplicate using the Lactate Pro (LacP) handheld analyzer and the StatProfile (StatP) reference instrument. The straight line indicates identity.
Figure 3. Difference plot and acceptance limits defined for analytic quality specification comparing L-lactate (LAC) measurements by the handheld Lactate Pro (LacP) analyzer with the StatProfile (StatP) reference instrument on 61 bovine blood samples. This plot represents the differences obtained between LAC concentrations measured by the handheld LacP and the StatP reference analyzer, plotted against LAC measured by the StatP. The acceptance limits for maximum allowable analytic bias are shown between the 2 lines (0 2.3 9 CVI 9 Mean). The coefficient of 2.3 was used due to the small number of samples used in the study (close to 70; therefore, there is a higher analytic bias than for higher sample numbers in which analytic bias is 0 1.96 9 CVI 9 Mean). Mean indicates the mean LAC measured in duplicate with LacP and StatP; CVI=inherent imprecision=√((CVLACPro)2 + (CVLACStatP)2) = 5.48%
Figure 1. Bland–Altman plot comparing whole blood L-lactate concentrations (LAC) measured by the Lactate Pro (LacP) and the StatProfile (StatP) reference instrument in 61 cattle. The solid line indicates the line of agreement between the results obtained by the 2 methods; the dashed lines represent the 95% confidence interval; the dotted line indicates the line of identity.
570
HCT data were available in 43 cattle with a mean (standard deviation [SD]) of 35% (7%), and a range from 22% – 51%. The HCT did not have a significant impact on the LAC using LacP (P = 0.39). There was a statistical trend toward a higher LAC concentration
Vet Clin Pathol 43/4 (2014) 567–572 ©2014 American Society for Veterinary Clinical Pathology
Buczinski et al
Figure 4. Classification plot of L-lactate concentration (LAC) data from the Lactate Pro (LacP) handheld analyzer and the reference StatProfile (StatP) analyzer on 61 bovine blood samples. The horizontal and vertical dotted lines represent the cutoff used to distinguish animals with a poor prognosis due to pneumonia (previously published cutoff of 4 mmol/L6). No sample is observed in the upper left quadrant indicating a specificity of 100% of the test (no false-positive with the LacP), the lower right quadrant contains false-negative samples (3 samples > 4 mmol/L with StatP, but < 4 mmol/L with LacP). The sensitivity of the LacP for that cutoff was SeLAC4 = 91.7%; 95% CI: 76.4–97.8% (33/36).
with the StatP analyzer in samples with a higher HCT (r = 0.26; P = .09).
Discussion Analysis of different biomarkers using handheld analyzers can be very valuable as they may be used cow-side. They could potentially serve in case management (eg, monitoring therapy and prognosis). Most of the handheld meters have been developed for human blood analysis; therefore, it is important to ensure that these units can be used accurately and safely in veterinary species. The results from the LacP handheld meter were highly correlated with those from a reference blood analyzer. However, the handheld analyzer underestimated LAC when compared with the reference analyzer, while it had an excellent intra-assay repeatability.21 The performance of the LacP is especially valuable for LAC < 10 mmol/L. At higher LAC values, performance and reproducibility of the LacP were inferior to the StatP analyzer. This is comparable to a previous study conducted in cats with the LacP.14 In cattle, however, previous studies did not assess handheld meters for LAC > 8 mmol/L10 or 9 mmol/L.6
Validation of handheld lactate meter in cattle
We used different statistical analyses to assess the handheld analyzer acceptability when compared with the reference method as previously recommended.15,20 Despite a high correlation between the 2 measurements, our data show that the 2 methods were not identical within the inherent analytic imprecision. This basically means that the value obtained in split samples by the handheld analyzer is not the same as the value obtained from the reference analyzer even if imprecision of the 2 methods is taken into account. The impact of this finding was partially evaluated using cutoffs previously defined in bovine medicine.5,6 Compared with the reference method, descriptive results of Se/Sp were > 85% except for SeLAC6. However, a relatively large 95% CI was found due to the limited number of cases in the present study with all lower 95% CI bound < 85% except for SpLAC6. For these reasons, we could not definitely state the clinical impact of this difference between both analyzers. The NLR for both cutoffs was compatible with a clinical use despite a wide 95% CI (which were all < 0.5 cutoff recognized as an indicator of small decrease of the likelihood of the outcome).16 The clinical acceptability of the LacP therefore indicates that it can be an interesting practical cow-side alternative providing an immediate result for LAC, although further studies in larger groups are required, especially because of the wide 95% CI found in the present study. Inversely to what has been found previously in horses9,22, HCT did not significantly influence LAC measurements with the LacP. Samples with high HCT analyzed with StatP tended to have higher LAC values compared with samples with lower HCT, which is in contrast to what has been found in equine studies. Previous studies in horses have shown that LAC values could be negatively influenced by HCT in animals with a HCT > 53%7 or 40%.22 A potential explanation for this observation could be the type of sample that was used (ie, whole blood vs plasma), which may lead to falsely decreased LAC in whole blood with a high HCT. This phenomenon is, however, not observed in people, even in samples with a very high HCT (> 65%) measured with a handheld analyzer.23 As both analyzers in this study used whole blood, we cannot draw final conclusions on the effect of HCT on LAC. Comparison of the LAC difference between whole blood and plasma samples would be necessary to assess this hypothesis. In conclusion, the handheld meter LacP is easy to use and provides precise and reproducible results compatible with a standard laboratory blood analyzer. It can therefore be used in cattle, especially for on-farm purpose. The LacP slightly underestimates LAC when compared with the reference instrument, which has
Vet Clin Pathol 43/4 (2014) 567–572 ©2014 American Society for Veterinary Clinical Pathology
571
Validation of handheld lactate meter in cattle
Buczinski et al
an impact on animals with LAC > 10 mmol/L. However, as most of the LAC thresholds proposed as prognostic cutoffs in bovine diseases are < 10 mmol/L4–6, the clinical importance of this phenomenon is likely small. As for other biomarkers, it is crucial to establish reference intervals and cutoffs for clinical decision that are specific for the respective instrument to avoid misdiagnosis and inadequate treatment decisions. Disclosure: The authors have indicated that they have no affiliations or financial involvement with any organization or entity with a financial interest in, or in financial competition with, the subject matter or materials discussed in this article.
dairy cows and calves. J Dairy Sci. 2012;95:6449– 6456. 11. Sloet van Oldruitenborgh-Oosterbaan MM, Van den Broek ET, Spierenburg AJ. Evaluation of the usefulness of the portable device Lactate Pro for measurement of lactate concentration in equine whole blood. J Vet Diagn Invest. 2008;20:83–85. 12. Delesalle C, Dewulf J, Lefebvre RA, et al. Determination of lactate concentrations in blood plasma and peritoneal fluid in horses with colic by an Accusport analyzer. J Vet Intern Med. 2007;21:293–301. 13. Thorneloe C, B edard C, Boysen S. Evaluation of a hand-held lactate analyzer in dogs. Can Vet J. 2007;48:283–288. 14. Acierno MJ, Johnson ME, Eddleman LA, Mitchell MA. Measuring statistical agreement between four point of care (POC) lactate meters and a laboratory blood analyzer in cats. J Feline Med Surg. 2008;10:110–114.
References 1. Pang DS, Boysen SR. Lactate in veterinary critical care: pathophysiology and management. J Am Anim Hosp Assoc. 2007;43:270–279.
15. Jensen AJ, Kjelgaard-Hansen M. Methods comparison in the clinical laboratory. Vet Clin Pathol. 2006;35: 276–286.
2. Lorenz I. D-lactic acidosis in calves. Vet J. 2009;179:197–203.
16. Halkin A, Reichman J, Schwaber M, Paltiel O, Brezis M. Likelihood ratios: getting diagnostic testing into perspective. Q J Med. 1998;91:247–258.
3. Owens FN, Secrist DS, Hill WJ, Gill DR. Acidosis in cattle: a review. J Anim Sci. 1998;76:275–286. 4. Figuereido MD, Nydam DV, Perkins GA, Mitchell HM, Divers TJ. Prognostic value of plasma L-lactate concentration measured cow-side with a portable clinical analyzer in Holstein dairy cattle with abomasal disorders. J Vet Intern Med. 2006;20:1463–1470. 5. Boulay G, Francoz D, Dor e E, et al. Preoperative value of a cow-side lactatemia measurement to predict negative outcome in Holstein dairy cattle with right abomasal disorders. J Dairy Sci. 2014;97:212–221. 6. Coghe J, Uystepruyst CH, Bureau F, Detilleux J, Art T, Lekeux P. Validation and prognostic value of plasma lactate measurement in bovine respiratory disease. Vet J. 2000;160:139–146. 7. Sorge UD, Kelton D, Staufenbiel R. Neonatal blood lactate concentration and calf mortality. Vet Rec. 2009;164:533–534. 8. Schulman ML, Nurton JP, Guthrie AJ. Use of the Accusport semi-automated analyser to determine blood lactate as an aid in the clinical assessment of horses with colic. J S Afr Vet Assoc. 2001;72:12–17. 9. Evans DL, Golland LC. Accuracy of Accusport for measurement of lactate concentration in equine blood or plasma. Equine Vet J. 1996;28:398–402. 10. Burfeind O, Heuwieser W. Validation of handheld meters to measure blood l-lactate concentration in
572
17. Dohoo I, Martin W, Stryhn H. Screening and diagnostic tests. In: Dohoo I, Martin W, Stryhn H, eds. Veterinary Epidemiologic Research. 2nd ed. Charlottetown: VER Publishing; 2010:91–133. 18. Seed P. Comparing several methods of measuring the same quality. Stata Technical Bulletin. 2000;10:2–9. 19. Bland JM, Altman DG. Statistical methods for assessing agreement between two methods of clinical measurement. Lancet. 1986;1:307–310. 20. Hyltoft-Petersen P, St€ ockl D, Blaajberg O, et al. Graphical interpretation of analytical data from comparison of a field method with a reference. Clin Chem. 1997;43:2039–2046. 21. Guidance for Industry: Bioanalytical Method Validation, US, 2001. Department of Health and Human Services, Food and Drug Administration, Center for Drug Evaluation and Research (CDER), Center for Veterinary Medicine (CVM) 2001, May, 5. 22. Tennent-Brown BS, Wilkins PA, Lindborg S, Russel G, Boston RC. Assessment of a point-of-care lactate monitoring emergency admission on adult horses to a referral hospital. J Vet Intern Med. 2007;21: 1090–1098. 23. Gambke B, Berg A, Fabian K, et al. Multicenter evaluation of a portable system for determining blood lactate. J Lab Med. 1997;21:250–256.
Vet Clin Pathol 43/4 (2014) 567–572 ©2014 American Society for Veterinary Clinical Pathology
ORIGINAL RESEARCH
Validation of the handheld Lactate-Pro analyzer for measurement of blood L-lactate concentration in cattle bastien Buczinski, Elizabeth Dore , Guillaume Boulay, David Francoz Se al, Montre al, QC, Canada Department of Clinical Sciences, Faculty of Veterinary Medicine, University of Montre
Key Words Biochemistry, bovine, dairy, handheld analyzer, lactatemia Correspondence S. Buczinski, Faculty of Veterinary Medicine, University of Montreal, CP5000, SaintHyacinthe, QC Montreal J2S 7C6, Canada E-mail: [email protected] DOI:10.1111/vcp.12185
Background: Blood L-lactate concentration (LAC) can be used for various diagnostic purposes in cattle. As multiple handheld analyzers for LAC exist, it is important to validate their use in cattle in comparison with reference laboratory blood analyzers. Objectives: The objectives of this study were to validate the handheld Lactate Pro meter (LacP) including reproducibility, and compare the measurements with the StatProfile (StatP) as a gold standard. In addition, diagnostic sensitivity and specificity, and the impact of HCT on LAC measured by both analyzers were assessed. Methods: A cohort of 64 cattle with acute medical and surgical conditions was studied. Whole blood samples in heparin lithium tubes were analyzed upon arrival with both StatP and LacP. Twenty-three samples were immediately retested to assess intra-assay coefficient of variation (CV). The HCT values were also recorded. Results: The LAC using LacP was highly correlated with the StatP (r = 0.9736 [95% confidence interval [CI]: 0.9562–0.9841]). The LacP underestimated LAC (mean difference:0.9 mmol/L, 95% CI:3.1 mmol/ L to 1.3 mmol/L). The intra-assay CV was excellent (4.77%). No significant correlation was observed between LacP or StatP and HCT (P = .39 and .09, respectively). Sensitivity and specificity for LacP were 91.7% (95% CI: 76.4–97.8%) and 100% (83.4–100%, cutoff of 4 mmol/L), and 78.6% (58.5–90.9%) and 100% (87.0–100%, cutoff of 6 mmol/L). Conclusions: The LacP handheld lactate meter can be used safely and reliably cow-side, although it underestimates LAC value when compared with a standard laboratory analyzer especially for LAC ≥ 10.0 mmol/L. The LAC value was not influenced by HCT in this study.
Introduction L-lactate is a sensitive marker of tissue perfusion as it is mainly produced during anaerobic glycolysis. Blood lactate may increase in various conditions including shock secondary to hypovolemia, endotoxemia, or heart failure, and is frequently used in diagnostic workups in human and veterinary medicine.1 By contrast to L-lactate which is produced by the mammalian cell, D-lactate, the other enantiomeric form of lactate, is produced by bacterial fermentation in the gastrointestinal tract of cattle.2 D-lactic acidosis has been implicated in various syndromes in adult and neonate cattle.2,3 All handheld analyzers of lactate assess L-lactate. A
rapid point-of-care analysis system for D-lactate assessment is currently not available. In cattle, rising blood L-lactate concentration (LAC) has been correlated with an unfavorable prognosis in cases of abomasal volvulus or right displaced abomasum4,5, or bovine respiratory disease.6 Specifically, in beef calves with bronchopneumonia6, LAC > 4 mmol/L was a reliable prognostic marker for animal death within 24 hours. We recently found that LAC > 6 mmol/L was associated with higher risk of poor outcome in the 30 days postsurgery in dairy cows with right abomasal dilation/volvulus.5 LAC has also been used for assessing the severity of neonatal calf distress shortly after birth.7 In horses, several studies have shown that LAC in whole blood can be underestimated
Vet Clin Pathol 43/4 (2014) 567–572 ©2014 American Society for Veterinary Clinical Pathology
567
Validation of handheld lactate meter in cattle
Buczinski et al
when measured using handheld analyzers in the presence of increased HCT.8,9 To date, scarce information is available concerning the correlation between HCT and LAC in cattle. One of the most important advantages of the handheld analyzers is that they can quickly assess LAC in a farm setting allowing more rapid and adequate patient management. Different handheld blood L-lactate analyzers are available for the clinician.4,6,10–12 Validation of these portative meters, including the determination of bias in comparison with standard chemistry analyzers, is of crucial importance particularly if LAC is used for prognostic assessments. The LactatePro (LacP) is a handheld L-lactate analyzer that gives a rapid result of LAC in the range from 0.8 to 23.3 mmol/L with only one drop of blood, and can be used easily at the farm. To date, its performance has been shown to be comparable to standard biochemistry analyzers in dogs13, cats14, and horses.11 However, the LacP has not been validated for cattle blood so far. The main objective of this study was to validate this device to measure LAC in cattle. Our hypothesis was that LAC is accurately estimated in cattle by the LacP analyzer in comparison with the StatProfile Critical Care Xpress (StatP) analyzer as a gold standard. A second objective was to assess the impact of HCT on LAC measurement by both analyzers.
Materials and Methods The study protocol was accepted by the ethical committee of the Faculty of veterinary medicine, University of Montr eal. In this prospective study, cows and calves admitted to the veterinary teaching hospital with acute medical and surgical conditions, including gastrointestinal diseases, various degrees of shock, or severe infectious processes, were eligible for the study. This study population was chosen (1) because of its anticipated variation of LAC values secondary to various degrees of hypoperfusion, shock, and infection, and (2) because the initial diagnostic workup of these animals routinely includes measurement of blood gas, pH, electrolytes, and L-lactate using a heparinized blood sample collected upon arrival. Gold standard measurements of LAC were performed using the StatProfile Critical Care Xpress (Nova Biomedical Corporation, Waltham, MA, USA). This analyzer uses an amperometric reaction to determine LAC. A whole blood sample of 130 lL is exposed to an electrode embedded with lactate oxydase. The lactate is converted to pyruvate and hydrogen peroxide. Hydrogen peroxide is oxidized
568
generating a current proportional to the LAC. This analyzer has an analytic range of 0.3–20.0 mmol/L. The data obtained were plotted in comparison with the values from the handheld device.13,15 The LactatePro (Arkray, Kyoto, Japan) is a portative handheld analyzer based on an amperometric method using an enzymatic reaction. The L-lactate reacts with the lactate oxydase, which reduces potassium ferricyanide to potassium ferrocyanide. The potassium ferrocyanide is then oxidized to potassium ferricyanide and produces an electric current proportional to lactate concentration of the sample. The result is displayed within 60 s after a drop of heparinized blood has been placed on the strip.
Estimate of random error for both methods To estimate the random error of analysis of both methods, a second analysis in both analyzers within 5 min allowed the assessment of the intra-assay coefficient of variation of both methods (CVLacP and CVStatP).
Clinical performance of the Lactate-Pro analyzer The efficiency of nonsurvival prediction by the LacP was determined for 2 different clinical cutoffs at 4 mmol/L for bronchopneumonia in calves6, and 6 mmol/L for right abomasal disorders in cows5 (determined by StatP analyzer as a gold standard). These 2 different cutoffs were chosen because of their clinical relevance in 2 major syndromes in bovine medicine. The clinical performance of the handheld analyzer was assessed evaluating sensitivity (Se) and specificity (Sp) of nonsurvivors detection at both cutoffs (4 or 6 mmol/L, respectively), and was designated SeLAC4, SpLAC4, SeLAC6, and SpLAC6. The formulae used were: (1) Sensitivity: SeLACCutoff (4 or 6 mmol/L, respectively) = number of samples with LAC using LacP ≥ 4 or 6 mmol/L, respectively/number of samples with LAC using StatP ≥ 4 or 6 mmol/L, respectively; (2) Specificity: SpLACCutoff (4 or 6 mmol/L, respectively) = number of samples with LAC using LacP < 4 or 6 mmol/L, respectively)/number of samples with LAC using StatP < 4 or 6 mmol/L, respectively. We defined acceptability criteria for the Se and Sp at a theoretical level of 85%. This arbitrary threshold was established based on mean values of different diagnostic tests used in human medicine.16 The 95% confidence intervals (CI) of these parameters were
Vet Clin Pathol 43/4 (2014) 567–572 ©2014 American Society for Veterinary Clinical Pathology
Buczinski et al
determined using a binomial distribution modeling as follows17: p 95% CI Se ¼ SeLAC4 or 6 1:96 ððSeLAC4 or 6 ð1 SeLAC4 or 6 Þ=nÞ; and p 95% CI Sp ¼ SpLAC4 or 6 1:96 ððSpLAC4 or 6 ð1 SpLAC4 or 6 Þ=nÞ; where SeLAC4 or 6 and SpLAC4 or 6 are the proportions found based on the contingency table and n, the total number of cases (n = 61 in this study). Due to the possible implication of this diagnostic test on individual management of patients, we calculated the negative and positive likelihood ratio (NLR and PLR, respectively). The NLR is defined as the number of true-positive patients (defined with the gold standard [StatP]) with a negative test result divided by the number of true-negative patients with a negative test result. The PLR represent the number of true-positive patients with a positive test result divided by the number of true-negative patients with a positive test result. One of the main interests of these parameters is that they do not depend on the prevalence of the outcome, but only on test Se and Sp, by contrast to predictive values.16 We aimed for a test that would have a good NLR (≤ 0.2)16 adequately identifying cows or calves with a favorable outcome (ie, survival). The NLR calculation was obtained from the formula: NLR = (1Se)/Sp as well as 95% CI.17 The PLR was also calculated using the formula PLR = Se/(1Sp) as an indicator of the odds ratio of nonsurvival based on the LacP analyzer classification. Where available, the HCT was determined by a hematology analyzer (Cell Dyn 3500, Abbott Laboratories, Mississauga, ON, USA) to assess its impact on the LAC values.
Statistical analysis All analyses were performed using commercial statistical software (SAS 9.2, Cary, NC, USA) including the graphics (MedCalc Software bvba, version 12.3.0, Mariakerke, Belgium). The relation between LAC determined by LacP and the StatP was determined using the Pearson correlation coefficient (r). A t-test was performed to determine if r was significantly different from 0. As comparison of different measurements using the Pearson correlation coefficient may have some limitations18, the agreement between the 2 methods was analyzed using the Bland–Altman analysis.19 The constant bias between the 2 methods and
Validation of handheld lactate meter in cattle
their limits of agreement (bias 2 SD) were also determined. The intra-assay coefficients of variation (CV) of both analyzers were calculated and expressed as a mean percentage. The combined inherent imprecision (CVI) was calculated using the formula for single measurements: p CVI ¼ ððCVLacP Þ2 þ ðCVStatP Þ2 Þ: The acceptability based on inherent imprecision of both tests was assessed using the mean value of both analytic methods and the difference between both methods, as well as both values corresponding to the 95% of the differences as previously recommended.20 As only 61 cases were used for this study, we expected that 95% of the differences between both methods would lie between (0 z 9 CVI 9 Mean of the 2 methods). The z value of 2.3 was used rather than the classical value of z = 1.96 for normally distributed data as previously recommended for data sets with < 70 samples.15 To assess the impact of HCT on the LAC measured by LacP and StatP analyzers, Pearson correlation coefficients were calculated and their statistical significance was determined as previously mentioned. The level of significance was set at P < .05.
Results A total of 64 cattle were eligible for inclusion in the study during the period from January 2011 to January 2012. The LAC range with the StatP analyzer was 0.4 mmol/L–> 20.0 mmol/L, and with the LacP analyzer < 0.8 mmol/L–19.9 mmol/L. Three animals were excluded from further analysis because they were out of range on one of the analyzers (2 animals with LAC < 0.8 mmol/L with LacP analyzer, and one with LAC > 20.0 mmol/L with StatP analyzer). Of the remaining 61 cattle, 45 had different acute gastrointestinal problems, 9 were downer-cows, one had bronchopneumonia, and 6 had no definitive diagnosis upon admission. The LAC using LacP analyzer ranged from 0.8 – 14.9 mmol/L (median: 4.9 mmol/L), and the LAC using StatP analyzer ranged from 0.6 mmol/L – 17.0 mmol/L (median: 5.7 mmol/L). The LAC from the LacP was highly correlated with LAC from the StatP (r = 0.9736, 95% CI: 0.9562– 0.9841; P < .01). The Bland–Altman analysis showed an excellent agreement between the reference method and the LacP when LAC was < 10 mmol/L (Figure 1). However, the LacP underestimated LAC compared
Vet Clin Pathol 43/4 (2014) 567–572 ©2014 American Society for Veterinary Clinical Pathology
569
Validation of handheld lactate meter in cattle
Buczinski et al
with the StatP analyzer (mean bias: 0.9 mmol/L, 95% CI from 3.1 mmol/L to 1.3 mmol/L). Differences between the 2 methods (LAC using LacP)-(LAC using StatP) ranged from 5.5 mmol/L to 0.4 mmol/L (median = 0.6 mmol/L). Twenty-three samples (from 0.9 mmol/ L – 12.6 mmol/L; median 3.6 mmol/L) were analyzed in duplicate with the LacP analyzer (Figure 2) and the StatP analyzer. The intra-assay CV for the LacP analyzer and the StatP analyzer was 4.77% and 2.69%, respectively. The inherent imprecision of both methods according to the formula (CVI = √((CVLacP)2 + (CVStatP)2) was 5.48%. Twenty-nine of 61 differences (48%) between the 2 tests were included in the inherent imprecision ranges (0 2.3 9 CVI 9 mean) (Figure 3). As less than 95% of the differences between both methods were within the imprecision ranges, this showed that the 2 methods were not identical and in fact reported different results for LAC in the same sample. For the 4 mmol/L cutoff, the LacP analyzer had a SeLAC4 of 91.7% (33 of 36 cases; 95% CI: 76.4–97.8%) and a SpLAC4 of 100% (25 of 25 cases; 95% CI: 83.4– 100%), and for the 6 mmol/L cutoff, SeLAC6 was 78.6% (22 of 28 cases; 95% CI: 58.5–90.9%) and SpLAC6 was 100% (33 of 33 cases; 95% CI: 87.0–100%) (Figure 4). The NLRLAC4 and NLRLAC6 were 0.08 (95% CI: 0.02–0.25) and 0.21 (95% CI: 0.11–0.44), respectively. As both SpLAC4 and SpLAC6 were 100%, it was not possible to calculate PLR.
Figure 2. Correlation of mean whole blood L-lactate concentrations (LAC) from 23 cattle measured in duplicate using the Lactate Pro (LacP) handheld analyzer and the StatProfile (StatP) reference instrument. The straight line indicates identity.
Figure 3. Difference plot and acceptance limits defined for analytic quality specification comparing L-lactate (LAC) measurements by the handheld Lactate Pro (LacP) analyzer with the StatProfile (StatP) reference instrument on 61 bovine blood samples. This plot represents the differences obtained between LAC concentrations measured by the handheld LacP and the StatP reference analyzer, plotted against LAC measured by the StatP. The acceptance limits for maximum allowable analytic bias are shown between the 2 lines (0 2.3 9 CVI 9 Mean). The coefficient of 2.3 was used due to the small number of samples used in the study (close to 70; therefore, there is a higher analytic bias than for higher sample numbers in which analytic bias is 0 1.96 9 CVI 9 Mean). Mean indicates the mean LAC measured in duplicate with LacP and StatP; CVI=inherent imprecision=√((CVLACPro)2 + (CVLACStatP)2) = 5.48%
Figure 1. Bland–Altman plot comparing whole blood L-lactate concentrations (LAC) measured by the Lactate Pro (LacP) and the StatProfile (StatP) reference instrument in 61 cattle. The solid line indicates the line of agreement between the results obtained by the 2 methods; the dashed lines represent the 95% confidence interval; the dotted line indicates the line of identity.
570
HCT data were available in 43 cattle with a mean (standard deviation [SD]) of 35% (7%), and a range from 22% – 51%. The HCT did not have a significant impact on the LAC using LacP (P = 0.39). There was a statistical trend toward a higher LAC concentration
Vet Clin Pathol 43/4 (2014) 567–572 ©2014 American Society for Veterinary Clinical Pathology
Buczinski et al
Figure 4. Classification plot of L-lactate concentration (LAC) data from the Lactate Pro (LacP) handheld analyzer and the reference StatProfile (StatP) analyzer on 61 bovine blood samples. The horizontal and vertical dotted lines represent the cutoff used to distinguish animals with a poor prognosis due to pneumonia (previously published cutoff of 4 mmol/L6). No sample is observed in the upper left quadrant indicating a specificity of 100% of the test (no false-positive with the LacP), the lower right quadrant contains false-negative samples (3 samples > 4 mmol/L with StatP, but < 4 mmol/L with LacP). The sensitivity of the LacP for that cutoff was SeLAC4 = 91.7%; 95% CI: 76.4–97.8% (33/36).
with the StatP analyzer in samples with a higher HCT (r = 0.26; P = .09).
Discussion Analysis of different biomarkers using handheld analyzers can be very valuable as they may be used cow-side. They could potentially serve in case management (eg, monitoring therapy and prognosis). Most of the handheld meters have been developed for human blood analysis; therefore, it is important to ensure that these units can be used accurately and safely in veterinary species. The results from the LacP handheld meter were highly correlated with those from a reference blood analyzer. However, the handheld analyzer underestimated LAC when compared with the reference analyzer, while it had an excellent intra-assay repeatability.21 The performance of the LacP is especially valuable for LAC < 10 mmol/L. At higher LAC values, performance and reproducibility of the LacP were inferior to the StatP analyzer. This is comparable to a previous study conducted in cats with the LacP.14 In cattle, however, previous studies did not assess handheld meters for LAC > 8 mmol/L10 or 9 mmol/L.6
Validation of handheld lactate meter in cattle
We used different statistical analyses to assess the handheld analyzer acceptability when compared with the reference method as previously recommended.15,20 Despite a high correlation between the 2 measurements, our data show that the 2 methods were not identical within the inherent analytic imprecision. This basically means that the value obtained in split samples by the handheld analyzer is not the same as the value obtained from the reference analyzer even if imprecision of the 2 methods is taken into account. The impact of this finding was partially evaluated using cutoffs previously defined in bovine medicine.5,6 Compared with the reference method, descriptive results of Se/Sp were > 85% except for SeLAC6. However, a relatively large 95% CI was found due to the limited number of cases in the present study with all lower 95% CI bound < 85% except for SpLAC6. For these reasons, we could not definitely state the clinical impact of this difference between both analyzers. The NLR for both cutoffs was compatible with a clinical use despite a wide 95% CI (which were all < 0.5 cutoff recognized as an indicator of small decrease of the likelihood of the outcome).16 The clinical acceptability of the LacP therefore indicates that it can be an interesting practical cow-side alternative providing an immediate result for LAC, although further studies in larger groups are required, especially because of the wide 95% CI found in the present study. Inversely to what has been found previously in horses9,22, HCT did not significantly influence LAC measurements with the LacP. Samples with high HCT analyzed with StatP tended to have higher LAC values compared with samples with lower HCT, which is in contrast to what has been found in equine studies. Previous studies in horses have shown that LAC values could be negatively influenced by HCT in animals with a HCT > 53%7 or 40%.22 A potential explanation for this observation could be the type of sample that was used (ie, whole blood vs plasma), which may lead to falsely decreased LAC in whole blood with a high HCT. This phenomenon is, however, not observed in people, even in samples with a very high HCT (> 65%) measured with a handheld analyzer.23 As both analyzers in this study used whole blood, we cannot draw final conclusions on the effect of HCT on LAC. Comparison of the LAC difference between whole blood and plasma samples would be necessary to assess this hypothesis. In conclusion, the handheld meter LacP is easy to use and provides precise and reproducible results compatible with a standard laboratory blood analyzer. It can therefore be used in cattle, especially for on-farm purpose. The LacP slightly underestimates LAC when compared with the reference instrument, which has
Vet Clin Pathol 43/4 (2014) 567–572 ©2014 American Society for Veterinary Clinical Pathology
571
Validation of handheld lactate meter in cattle
Buczinski et al
an impact on animals with LAC > 10 mmol/L. However, as most of the LAC thresholds proposed as prognostic cutoffs in bovine diseases are < 10 mmol/L4–6, the clinical importance of this phenomenon is likely small. As for other biomarkers, it is crucial to establish reference intervals and cutoffs for clinical decision that are specific for the respective instrument to avoid misdiagnosis and inadequate treatment decisions. Disclosure: The authors have indicated that they have no affiliations or financial involvement with any organization or entity with a financial interest in, or in financial competition with, the subject matter or materials discussed in this article.
dairy cows and calves. J Dairy Sci. 2012;95:6449– 6456. 11. Sloet van Oldruitenborgh-Oosterbaan MM, Van den Broek ET, Spierenburg AJ. Evaluation of the usefulness of the portable device Lactate Pro for measurement of lactate concentration in equine whole blood. J Vet Diagn Invest. 2008;20:83–85. 12. Delesalle C, Dewulf J, Lefebvre RA, et al. Determination of lactate concentrations in blood plasma and peritoneal fluid in horses with colic by an Accusport analyzer. J Vet Intern Med. 2007;21:293–301. 13. Thorneloe C, B edard C, Boysen S. Evaluation of a hand-held lactate analyzer in dogs. Can Vet J. 2007;48:283–288. 14. Acierno MJ, Johnson ME, Eddleman LA, Mitchell MA. Measuring statistical agreement between four point of care (POC) lactate meters and a laboratory blood analyzer in cats. J Feline Med Surg. 2008;10:110–114.
References 1. Pang DS, Boysen SR. Lactate in veterinary critical care: pathophysiology and management. J Am Anim Hosp Assoc. 2007;43:270–279.
15. Jensen AJ, Kjelgaard-Hansen M. Methods comparison in the clinical laboratory. Vet Clin Pathol. 2006;35: 276–286.
2. Lorenz I. D-lactic acidosis in calves. Vet J. 2009;179:197–203.
16. Halkin A, Reichman J, Schwaber M, Paltiel O, Brezis M. Likelihood ratios: getting diagnostic testing into perspective. Q J Med. 1998;91:247–258.
3. Owens FN, Secrist DS, Hill WJ, Gill DR. Acidosis in cattle: a review. J Anim Sci. 1998;76:275–286. 4. Figuereido MD, Nydam DV, Perkins GA, Mitchell HM, Divers TJ. Prognostic value of plasma L-lactate concentration measured cow-side with a portable clinical analyzer in Holstein dairy cattle with abomasal disorders. J Vet Intern Med. 2006;20:1463–1470. 5. Boulay G, Francoz D, Dor e E, et al. Preoperative value of a cow-side lactatemia measurement to predict negative outcome in Holstein dairy cattle with right abomasal disorders. J Dairy Sci. 2014;97:212–221. 6. Coghe J, Uystepruyst CH, Bureau F, Detilleux J, Art T, Lekeux P. Validation and prognostic value of plasma lactate measurement in bovine respiratory disease. Vet J. 2000;160:139–146. 7. Sorge UD, Kelton D, Staufenbiel R. Neonatal blood lactate concentration and calf mortality. Vet Rec. 2009;164:533–534. 8. Schulman ML, Nurton JP, Guthrie AJ. Use of the Accusport semi-automated analyser to determine blood lactate as an aid in the clinical assessment of horses with colic. J S Afr Vet Assoc. 2001;72:12–17. 9. Evans DL, Golland LC. Accuracy of Accusport for measurement of lactate concentration in equine blood or plasma. Equine Vet J. 1996;28:398–402. 10. Burfeind O, Heuwieser W. Validation of handheld meters to measure blood l-lactate concentration in
572
17. Dohoo I, Martin W, Stryhn H. Screening and diagnostic tests. In: Dohoo I, Martin W, Stryhn H, eds. Veterinary Epidemiologic Research. 2nd ed. Charlottetown: VER Publishing; 2010:91–133. 18. Seed P. Comparing several methods of measuring the same quality. Stata Technical Bulletin. 2000;10:2–9. 19. Bland JM, Altman DG. Statistical methods for assessing agreement between two methods of clinical measurement. Lancet. 1986;1:307–310. 20. Hyltoft-Petersen P, St€ ockl D, Blaajberg O, et al. Graphical interpretation of analytical data from comparison of a field method with a reference. Clin Chem. 1997;43:2039–2046. 21. Guidance for Industry: Bioanalytical Method Validation, US, 2001. Department of Health and Human Services, Food and Drug Administration, Center for Drug Evaluation and Research (CDER), Center for Veterinary Medicine (CVM) 2001, May, 5. 22. Tennent-Brown BS, Wilkins PA, Lindborg S, Russel G, Boston RC. Assessment of a point-of-care lactate monitoring emergency admission on adult horses to a referral hospital. J Vet Intern Med. 2007;21: 1090–1098. 23. Gambke B, Berg A, Fabian K, et al. Multicenter evaluation of a portable system for determining blood lactate. J Lab Med. 1997;21:250–256.
Vet Clin Pathol 43/4 (2014) 567–572 ©2014 American Society for Veterinary Clinical Pathology
