This article was downloaded by: [New York University] On: 09 June 2015, At: 03:58 Publisher: Taylor & Francis Informa Ltd Registered in England and Wales Registered Number: 1072954 Registered office: Mortimer House, 37-41 Mortimer Street, London W1T 3JH, UK
Veterinary Quarterly Publication details, including instructions for authors and subscription information: http://www.tandfonline.com/loi/tveq20
Critical thresholds of liver function parameters for ketosis prediction in dairy cows using receiver operating characteristic (ROC) analysis a
a
a
a
a
a
ab
Yuhang Sun , Bo Wang , Shi Shu , Hongyou Zhang , Chuang Xu , Ling Wu & Cheng Xia a
Department of College of Animal Science and Veterinary Medicine, Heilongjiang BaYi Agriculture University, Daqing 163319, China b
Department of Synergetic Innovation Center of Food Safety and Nutrition, Northeast Agricultural University, Harbin 150030, China Published online: 02 Apr 2015.
Click for updates To cite this article: Yuhang Sun, Bo Wang, Shi Shu, Hongyou Zhang, Chuang Xu, Ling Wu & Cheng Xia (2015): Critical thresholds of liver function parameters for ketosis prediction in dairy cows using receiver operating characteristic (ROC) analysis, Veterinary Quarterly, DOI: 10.1080/01652176.2015.1028657 To link to this article: http://dx.doi.org/10.1080/01652176.2015.1028657
PLEASE SCROLL DOWN FOR ARTICLE Taylor & Francis makes every effort to ensure the accuracy of all the information (the “Content”) contained in the publications on our platform. However, Taylor & Francis, our agents, and our licensors make no representations or warranties whatsoever as to the accuracy, completeness, or suitability for any purpose of the Content. Any opinions and views expressed in this publication are the opinions and views of the authors, and are not the views of or endorsed by Taylor & Francis. The accuracy of the Content should not be relied upon and should be independently verified with primary sources of information. Taylor and Francis shall not be liable for any losses, actions, claims, proceedings, demands, costs, expenses, damages, and other liabilities whatsoever or howsoever caused arising directly or indirectly in connection with, in relation to or arising out of the use of the Content. This article may be used for research, teaching, and private study purposes. Any substantial or systematic reproduction, redistribution, reselling, loan, sub-licensing, systematic supply, or distribution in any form to anyone is expressly forbidden. Terms & Conditions of access and use can be found at http:// www.tandfonline.com/page/terms-and-conditions
Veterinary Quarterly, 2015 http://dx.doi.org/10.1080/01652176.2015.1028657
SHORT COMMUNICATION Critical thresholds of liver function parameters for ketosis prediction in dairy cows using receiver operating characteristic (ROC) analysis Yuhang Suna, Bo Wanga,†, Shi Shua, Hongyou Zhanga, Chuang Xua, Ling Wua and Cheng Xiaa,b* a
Department of College of Animal Science and Veterinary Medicine, Heilongjiang BaYi Agriculture University, Daqing 163319, China; b Department of Synergetic Innovation Center of Food Safety and Nutrition, Northeast Agricultural University, Harbin 150030, China
Downloaded by [New York University] at 03:58 09 June 2015
(Received 28 May 2014; accepted 9 March 2015) Background: Fatty liver syndrome and ketosis are important metabolic disorders in high-producing cows during early lactation with fatty liver usually preceding ketosis. To date, parameters for early prediction of the risk of ketosis have not been investigated in China. Objective: To determine the predictive value of some parameters on the risk of ketosis in China. Animals and methods: In a descriptive study, 48 control and 32 ketotic Holstein Friesian cows were randomly selected from one farm with a serum b-hydroxybutyrate (BHBA) concentration of 1.20 mmol/L as cutoff point. The risk prediction thresholds for ketosis were determined by receiver operating characteristic (ROC) analysis. Results: In line with a high BHBA concentration, blood glucose concentration was significantly lower in ketotic cows compared to control animals (2.77 § 0.24 versus 3.34 § 0.03 mmol/L; P D 0.02). Thresholds were more than 0.76 mmol/L for nonesterified fatty acids (NEFA, with 65% sensitivity and 92% specificity), more than 104 U/L for aspartate aminotransferase (AST, 74% and 85%, respectively), less than 140 U/L for cholinesterase (CHE, 75% and 59%, respectively), and more than 3.3 mmol/L for total bilirubin (TBIL, 58% and 83%, respectively). There were significant correlations between BHBA and glucose (R D 4.74), or CHE (R D 0.262), BHBA and NEFA (R D 0.520), or AST (R D 0.525), or TBIL (R D 0.278), or direct bilirubin (DBIL, R D 0.348). Conclusions: AST, CHE, TBIL and NEFA may be useful parameters for risk prediction of ketosis. Clinical importance: This study might be of value in addressing novel directions for future research on the connection between ketosis and liver dysfunction. Keywords: bovine; dairy cows; ketosis; liver function; risk prediction; threshold; ROC analysis
1. Introduction Fatty liver syndrome and ketosis are some important metabolic disorders in high-producing dairy cows especially during early lactation. When energy intake cannot meet the energy demand of the mammary gland for milk production and other metabolic functions, fat mobilization from adipose tissue rapidly increases. This releases a large amount of nonesterified fatty acids (NEFA) into the blood circulation, which are transported to the liver. Given that the oxidizing ability of the liver regarding NEFA in dairy cows is limited, excessive NEFA in the liver are converted into ketone bodies, or accumulated in hepatocytes in the form of triacylglycerol leading to ketosis and/or fatty liver syndrome (Rukkwamsuk et al. 1999, Gathercole et al. 2012). As a consequence, fatty liver syndrome is closely related to ketosis. Usually, the fatty liver syndrome is also associated with abnormalities in liver function, which result in reduction of NEFA oxidization in the tricarboxylic acid cycle and the synthesis of lipoprotein, and as a consequence the blood concentrations of ketone bodies increase rapidly and persistently. Ketosis is experimentally induced by fasting (Brumby et al. 1975) and by administration of 1,3-butanediol (Veenhuizen & co-workers 1991). In both cases, fatty
liver precedes ketosis. On the other hand, fatty liver is found in almost all cases of natural ketosis. As in fatty liver syndrome, overfeeding is epidemiologically suspected as a major causal factor for ketosis (Markusfeld 1985). However, it should be realized that in normal cows, triacylglycerol accumulation in the liver at the end of the dry period is physiological. A large concentration of ketone bodies in the blood may cause anorexia, or decreased feed intake, which worsens the negative energy balance further, to provoke severe ketosis and fatty liver. Affected cows are unhealthy and show poor production performance, including decreases in dry matter intake (DMI) and milk yield (MY) (Dohoo & Martin 1984, Dohoo et al. 1984), which cause financial losses and poor animal welfare. The concentrations of bile components (e.g. bilirubin, bile acids and cholesterylesters) and triglyceride (TG) increase in the plasma of dairy cows with fatty liver syndrome, while the concentrations of free cholesterol, citrate and glycogen decrease. Fatty liver syndrome is usually characteristic of cows with high blood concentrations of NEFA, butyrate and acetoacetic acid, which may be toxic to cells and reduce b-oxidation and gluconeogenesis in the liver (Ametaj et al. 2005). In addition to the increase
*Corresponding author. Email: [email protected] † Parallel first author: Bo Wang This article was originally published with errors. This version has been corrected. Please see Erratum (http://dx.doi.org/10.1080/ 01652176.2015.1040689). Ó 2015 Taylor & Francis
Downloaded by [New York University] at 03:58 09 June 2015
2
Y. Sun et al.
in blood NEFA concentrations, decreased concentrations in cholesterylesters and phospholipids are observed in cows with ketosis (Kronfeld 1965, Watson & Williams 1987, Katoh 2002). However, current detection of fatty liver is possible only by minor surgery and subsequent biopsy analysis (Bobe & co-workers 2005). Some blood parameters, such as total protein (TP), albumin (ALB), cholinesterase (CHE) and prothrombin time, are commonly considered to reflect the synthetic ability of the liver because they are only synthesized in hepatocytes and have a positive association with their synthetic ability. Once liver synthetic capability decreases, their values in plasma also decease. In addition, the activities of aspartate aminotransferase (AST or GOT), alanine aminotransferase (ALT or GPT), adenosine deaminase and lactic dehydrogenase (LDH) in plasma are often used as biomarkers that reflect whether hepatocytes are damaged and the degree of hepatocyte damage. When liver cell membranes are impaired or liver cell necrosis occurs, these enzymes are released into the blood. Therefore, assay of enzyme activity in plasma or serum can predict the degree of exposure in hepatic cells (Steen et al. 1997). ALT is the most sensitive target in the detection of acute liver damage and AST is a more sensitive indicator of the degree of exacerbated liver damage. When 1% of liver cells have undergone necrosis, the serum ALT activity doubles. However, the activity of AST increases continuously to exceed the value of ALT; this suggests that severe damage has occurred to liver parenchyma and is an indication of exacerbated illness (Giannini et al. 1999). Given that abnormalities of hepatic function may be observed in relation to excretion, secretion, transportation and detoxication, the plasma concentrations of total bilirubin (TBIL), direct bilirubin (DBIL), total bile acid and blood ammonia usually increase (Patel & Gores 1995). Among the available parameters, it is necessary to select a sensitive, specific, low-cost and simple target to assess liver function. It is accepted that the blood concentration of b-Hydroxybutyrate (BHBA) is the gold standard for the diagnosis of ketosis because blood BHBA has high sensitivity, specificity, accuracy and stability (Duffield 2000, Atkinson et al. 2004). A serum BHBA concentration above 1.2 mmol/L is commonly used as a cutoff point for detection of ketosis (LeBlanc 2010, Ospina et al. 2010). Previous reports on the critical thresholds for ketosis prediction have incorporated receiver operating characteristic (ROC) analysis (Ospina et al. 2010, Asl et al. 2011). Mulligan et al. suggested that milk data may be used to assess the energy balance in early lactation; other assessments include feed intake and refusal of intake, blood BHBA and NEFA concentrations 14 days prepartum, and on-farm approaches to control postpartum ketosis (Mulligan et al. 2006). Although liver dysfunction is closely associated with fatty liver and ketosis, it is unclear whether liver function parameters can predict ketosis. To date, the cutoff point of liver function parameters for early prediction of the risk of ketosis has not been investigated in China. Therefore, the objective of this study was to investigate the
relationship between liver function and ketosis in China using binary logistic regression analysis, and to determine the critical thresholds of liver function parameters for ketosis prediction using ROC curves in order to provide a scientific basis for the prevention of ketosis in dairy cows.
2. Materials and methods 2.1. Animals The study was carried out in a dairy herd comprising more than 3000 Holstein Friesian cows in total that was maintained in a free-stall housing using a complete dairy management software (Valley Agricultural Software, Tulare, CA, USA) system in Heilongjiang, China. The average calving interval of the herd was 444 days and the mean (305 days) MY 23.6 kg/day associated with 4.55% fat and 3.30% protein. From April to June 2012, 80 multiparous healthy cows and cows with ketosis were selected randomly generating 48 control cows and 32 ketotic cows; basic information and MY were recorded. The cows were considered to have ketosis (K) if they had serum BHBA concentrations >1.20 mmol/L, otherwise they were regarded as healthy controls (C) (Oetzel 2004). All the cows were fed a total mixed ration during the transition period, which consisted of 8.5 kg concentrated feed, 18.5 kg maize silage, 4 kg hay and 0.35 kg fat. Its nutritional levels were 57% dry matter (DM), 17% crude protein, 1.76 mcal/DM net energy for lactation, 5.6% fat, 38.5% neutral detergent fiber, 20.3% acid detergent fiber, 180 g Ca, and 100 g P, with a daily DMI of 19 kg. 2.2.
Sample collection
Blood samples were taken from the caudal vein into a plain tube and a tube containing EDTA (ethylene diamine tetraacetic acid) between 14 and 21 days postpartum, before the morning feed. EDTA blood samples were immediately centrifuged at 1400 g for 5 min at room temperature. The supernatants were aliquoted into Eppendorf tubes (1 ml plasma/tube) and stored at 80 C until analysis. All haemolytic samples were discarded. This study was carried out in strict accordance with the recommendations in the Guide for the Care and Use of Laboratory Animals of the National Institutes of Health, USA. All animal experiments were conducted according to the practices and standards approved by the Animal Welfare and Research Ethics Committee at Heilongjiang Bayi Agricultural University, China. 2.3. Laboratory analysis BHBA was measured by means of an enzyme-linked immunosorbent assay (Nanjing Senbeijia Biotechnology Company, Nanjing China); AST, CHE and ALT were detected by the velocity method (Beijing Strong Biotechnology Company, Beijing, China). NEFA were measured by an enzymic method (Nanjing Senbeijia Biotechnology Company, Nanjing China); glucose (Glc) was detected by
Veterinary Quarterly the glucose oxidase method; TBIL and DBIL were analyzed by the dyeline method (Shanghai Desai Biotechnology Company, Shanghai, China). TP and ALB were measured by a colorimetric method (ACCU-CHEK, Roche, Basel, Switzerland). All the plasma samples were analyzed using an automatic biochemical analyzer (HITACHI 7600, Kyoto, Japan).
Downloaded by [New York University] at 03:58 09 June 2015
2.4.
Statistical analysis
Significant analysis of several parameters: student’s t-test was used to analyze any significant differences in plasma concentrations of Glc, BHBA, NEFA, ALT, AST, CHE, TBIL, DBIL, indirect bilirubin (IBIL), TP, ALB, globulin (GLB) and MY between the healthy group and the ketotic group. The analyses were performed using SPSS software (Version 19.0, IBM, New York, USA). All data were presented as mean and standard error (SE) in Table 1, and P < 0.05 was considered to represent statistical significance. Pearson’s correlation coefficients: given that blood BHBA concentration has been accepted as gold standard for the diagnosis of ketosis and was also considered as an optimal herd monitoring biomarker (Oetzel 2004), Pearson’s correlation coefficients were utilized to reveal the correlations between BHBA and the parameters in Table 1. Regression analysis: subsequently, two logistic models were established to screen risk factors for negative energy balance associated with ketosis. Model 1 included Glc and NEFA to determine the causal relationship between energy balance and ketosis, and other risk factors for liver dysfunction included in Model 2 were AST and CHE, to confirm the causal relationship between liver function and ketosis.
Table 1. Mean (§ standard error) and P-value for milk yield and 12 blood biochemical parameters in the control and ketosis groups. Parameter Number MY (kg/day) BHBA (mmol/L) NEFAs (mmol/L) Glc (mmol/L) ALT (U/L) AST (U/L) CHE (U/L) TP (g/L) ALB (g/L) GLB (g/L) TBIL (mmol/L) DBIL (mmol/L) IBIL (mmol/L)
C
K
P-value
48 33.42 § 1.02 0.61 § 0.02 0.54 § 0.04 3.34 § 0.03 21.94 § 0.74 89.27 § 2.82 153.69 § 2.95 78.69 § 0.92 34.12 § 0.36 44.51 § 1.13 2.75 § 0.11 1.79 § 0.09 0.95 § 0.07
32 33.05 § 1.96 2.84 § 0.21 0.87 § 0.06 2.77 § 0.24 22.06 § 0.89 134.84 § 8.63 139.03 § 3.58 75.46 § 0.88 32.78 § 0.51 42.71 § 0.87 3.61 § 0.28 2.51 § 0.18 1.09 § 0.13
0.87 0.00 0.00 0.02 0.91 0.00 0.00 0.02 0.03 0.21 0.01 0.00 0.31
Note: K: ketosis group, C: control group; MY: milk yield, Glc: glucose, BHBA: b-hydroxybutyrate, NEFAs: nonesterified fatty acids, ALT: alanine aminotransferase, AST: aspartate aminotransferase, CHE: cholinesterase, TP: total protein, ALB: albumin, GLB: globulin, TBIL: total bilirubin, DBIL: direct bilirubin, IBIL: indirect bilirubin.
3
ROC analysis: finally, the cutoff points or critical thresholds of the risk factors were determined using ROC analysis, in which BHBA was included as a continuous outcome parameter. Furthermore, curves of sensitivity versus specificity were plotted using ROC analysis for NEFA, AST, CHE, DBIL, etc., to include all possible threshold values of the parameters assessed. Youden’s index was computed to determine the optimal thresholds of the selected parameters. The area under the curve (AUC) was used to assess the diagnostic precision of the eligible parameters (Gardner & Greiner 2006).
3. Results 3.1. Significance test The clinical data and biochemical parameters are presented in Table 1. The analysis included both subclinical cases and clinical cases, which may have caused the wide range observed in MY and the median values of the two groups to be similar. A box plot (Figure 1) is presented to demonstrate this phenomenon. 3.2.
Pearson’s correlation coefficients
In the 80 cows tested, there was a significant negative correlation between the plasma concentrations of BHBA and Glc (R D 4.74, P D 0.000), or CHE (R D 0.262, P D 0.019), and also a prominent positive correlation between the plasma concentrations of BHBA and NEFA (R D 0.520, P D 0.000), AST (R D 0.525, P D 0.000), TBIL (R D 0.278, P D 0.012) or DBIL (R D 0.348, P D 0.002). However, there was no significant correlation between the plasma concentrations of BHBA and ALT, TP, ALB, GLB or IBIL (P 0.05). 3.3.
Regression analysis
In the first logistic model, plasma Glc (P D 0.101) was removed and NEFA (P D 0.000) was retained, suggesting that an increase in NEFA can predict the rise in BHBA. In the second logistic model, plasma AST (P D 0.000), CHE
Figure 1. Box plot of milk yield in Holstein Friesian cows for control group (n D 48) and ketosis group (n D 32).
4
Y. Sun et al.
Table 2. Prediction point, sensitivity, specificity, respective SE, CLR and AUC for NEFAs, AST, CHE and TBIL (n D 48 control and 32 ketotic cows). Parameter NEFAs (mmol/L) AST (U/L) CHE (U/L) TBIL (mmol/L)
Prediction point
Sensitivity (%)
SE (sen.%)
Specificity (%)
SE (spe.%)
CLR
AUC
0.76 104.0 139.5 3.25
64.5 74.2 75.0 58.1
9.88 8.13 8.82 9.73
91.7 85.4 59.4 83.3
3.69 5.04 6.25 5.07
7.77 5.08 1.85 3.48
0.818, p D 0.000 0.849, p D 0.000 0.704, p D 0.002 0.714, p D 0.001
Note: SE: standard error, CLR: positive likelihood ratio; NEFAs: nonesterified fatty acids, AST: aspartate aminotransferase, CHE: cholinesterase, TBIL: total bilirubin, AUC: area under ROC curve.
Downloaded by [New York University] at 03:58 09 June 2015
(P D 0.001) and TBIL (P D 0.034) were retained and DBIL (P D 0.354) was eliminated. 3.4. ROC analysis The prediction thresholds, sensitivity, specificity, respective SE, positive likelihood ratios (CLR) and areas under the ROC curves for NEFA, AST, CHE and TBIL are presented in Table 2. The optimal prediction thresholds were determined by Youden’s index to be more than 0.76 mmol/L for NEFA, with 65% sensitivity and 92% specificity, more than 104 U/L for AST, with 74% sensitivity and 85% specificity, less than 140 U/L for CHE, with 75% sensitivity and 59% specificity, and more than 3.25 mmol/L for TBIL, with 58% sensitivity and 83% specificity. The area under the ROC curves for NEFA, AST, CHE and TBIL are shown in Figure 2. As CHE had a negative correlation with BHBA, the area above the ROC curve was selected. The area under (or above) the ROC curve was greater than 0.70 for all four parameters (Table 2). 4. Discussion As shown in Table 1, when cows with ketosis were compared with healthy controls, there were significant increases in mean plasma concentrations of BHBA, NEFA, AST, TBIL and DBIL (P < 0.01), obvious decreases in plasma CHE (P < 0.01), Glc, TP and ALB
Figure 2. Receiver operating characteristic (ROC) plots for NEFA, AST, CHE and TBIL for diagnosis of ketosis in Holstein Friesian cows (NEFAs: nonesterified fatty acids; AST: aspartate aminotransferase; CHE: cholinesterase; TBIL: total bilirubin).
(P < 0.05), but no significant change in plasma concentrations of IBIL, GLB, ALT and MY. In addition, in subsequent Pearson’s correlation coefficients analysis, there was a significant correlation between the plasma concentrations of BHBA and glucose, CHE, NEFA, AST, TBIL or DBIL. Blood ketone body and glucose concentrations are the main diagnostic parameters for ketosis. An increased blood ketone body concentration accompanied by a decreased blood glucose concentration, becomes the main risk factor for ketosis (Ingvartsen 2006) and the low glucose concentration reflects the negative energy balance and triggers lipolysis due to the associated low insulin concentration. BHBA is the primary component of blood ketone bodies in dairy cattle; it is relatively stable in whole blood, serum or plasma, both in vitro and in vivo (Custer et al. 1983). Thus, blood BHBA concentration has been accepted as gold standard for the diagnosis of ketosis and was also considered as an optimal herd monitoring biomarker (Oetzel 2004). As a result, in this study BHBA, glucose, NEFA, AST, TBIL, DBIL and CHE were considered to be associated with ketosis, and can be regarded as novel biomarkers for diagnosis of ketosis. However, it should be realized that blood concentrations of BHBA and NEFA in early lactation are higher than in mid-lactation. The prevalence of a high level of NEFA in early lactation has been reported to be 67%, but was only 17% in mid-lactation, whereas the prevalence of subclinical ketosis in early lactation may reach 41%, but in mid-lactation 28% (Gonzalez et al. 2011). In the subsequent regression analysis, NEFA was retained and glucose was removed. This is in accordance with the finding of Asl that NEFA has a predictive effect for ketosis but glucose does not (Asl et al. 2011). In the second logistic model, several parameters associated with liver function, namely AST, CHE and TBIL were retained. It implied that the plasma levels of NEFA, AST, CHE and TBIL may predict the risk of ketosis, and liver dysfunction may become an important risk factor for ketosis. In the first months of lactation, the plasma concentrations of NEFA and TBIL should decrease gradually, while total lipids should increase gradually in healthy dairy cows (Dale et al. 1979). In this study, the mean NEFA concentration was significantly different (P D 0.000) between the control and ketotic cows (0.54 and 0.87 mmol/L, respectively). The optimal threshold for NEFA was determined to be 0.76 mmol/L using ROC analysis, which is similar to the NEFA value of more than
Downloaded by [New York University] at 03:58 09 June 2015
Veterinary Quarterly 0.72 mmol/L reported by Ospina (Ospina et al. 2010). However, this result was higher than those of other experiments, such as the 0.4 mmol/L reported by Oetzel (Oetzel 2004) and 0.26 mmol/L by Asl (Asl et al. 2011). This phenomenon may be related to the high level of milk production (33 kg/day) and the high incidence of ketosis (40%) in the dairy herd studied herein. This suggests a need to adopt nutritional management and preventive measures to reduce the likelihood of ketosis. Plasma levels of ALT, AST and CHE, which are released from liver cells, can be used as important parameters that reflect whether hepatocyte damage has occurred, and the extent of such damage. When liver cell membranes are impaired or liver cell necrosis occurs, most of these enzymes are released into the blood. Assay of enzyme activity in plasma or serum can demonstrate the degree of damage to hepatic cells (Steen et al. 1997). Among clinical enzyme tests, AST and ALT may be used to investigate the type and degree of damage to liver cells. As mentioned before, ALT is the most sensitive target in the diagnosis of acute liver damage, whereas AST is more sensitive in reflecting the degree of damage (Kew 2000). The normal activity of AST in blood is 78 132 U/L in healthy cattle (Ingvartsen 2006). In this study, the mean activity of AST was 89 U/L in the control group. This result was close to that reported by Moallem (Moallem et al. 2004). AST exists primarily in mitochondria. When mild liver damage occurs, the increase of AST is lower than that of ALT, i.e. AST/ALT 0.05). So, the correlation between TP, ALB or GLB and liver function is not substantiated yet. In this study, blood TBIL, DBIL and IBIL were considered as indicators reflecting the capacity of the liver for excretion, secretion and detoxication. When hepatocytes are damaged, the hepatic capability for excretion, secretion and detoxication are reduced, and this results in the increase of the three parameters mentioned above. The TBIL (P D 0.007) and DBIL (P D 0.000) in the ketosis group were significantly higher than in the control group, but IBIL showed no difference between both groups. In the logistic regression analysis, only TBIL (P D 0.034) stayed in the model; DBIL (P D 0.703) did not have any predictive value for ketosis. The prediction threshold of TBIL was more than 3.3 mmol/L, with 58% sensitivity and 83% specificity. Therefore, a plasma concentration of TBIL higher than 3.3 mmol/L suggests abnormal liver function, and can also be considered as a predictor for a high risk of ketosis. In conclusion, the results of this study showed that the prevalence of ketosis, including both clinical and subclinical ketosis, in dairy cattle on a farm in Heilongjiang Province, China, is significant. For their greater sensitivity and specificity, the values of NEFA, AST, CHE and TBIL can be used during early lactation for prediction of the risk of ketosis and liver dysfunction, which will not only provide new references for prediction and diagnosis of ketosis, but also finds novel directions for research of the connection between ketosis and liver dysfunction. Furthermore, to prevent the economic loss due to ketosis associated with liver dysfunction, early treatment of affected cows is necessary. Prediction and prevention of ketosis must be achieved through monitoring of liver function and proper nutritional programs for dry cows and those in early lactation. Disclosure statement No potential conflict of interest was reported by the authors.
Funding The authors thank the National Science Foundation Committee [31001094]; the China and National Science and Technology Project ‘Twelfth Five-Year’ in rural areas [2012BAD12B05-2], [2012BAD12B03-2]; the National Science and Technology Support Program [2013BAD21B01]; the National Science Foundation Committee [31672625]; China Spark Program [2012GA670001] and NCET [1252-NCET-003] for the financial support.
6
Y. Sun et al.
Downloaded by [New York University] at 03:58 09 June 2015
References Ametaj B, Bradford B, Bobe G, Nafikov R, Lu Y, Young J, Beitz D. 2005. Strong relationships between mediators of the acute phase response and fatty liver in dairy cows. Can J Animal Sci. 85:165 175. Asl AN, Nazifi S, Ghasrodashti AR, Olyaee A. 2011. Prevalence of subclinical ketosis in dairy cattle in the Southwestern Iran and detection of cutoff point for NEFA and glucose concentrations for diagnosis of subclinical ketosis. Preventive Veterinary Med. 100:38 43. Atkinson SJ, Hosford MA, Molitoris BA. 2004. Mechanism of actin polymerization in cellular ATP depletion. J Biol Chem. 279:5194 5199. Brumby PE, Anderson M, Tuckley B, Storry JE, Hibbit KG. 1975. Lipid metabolism in the cow during starvationinduced ketosis. Biochem J. 146(3):609 615. Custer E, Myers J, Poffenbarger P, Schoen I. 1983. The storage stability of 3-hydroxybutyrate in serum, plasma, and whole blood. Am J Clin Pathol. 80:375 380. Dale H, Vik-Mo L, Fjellheim P. 1979. A field survey of fat mobilization and liver function of dairy cows during early lactation. Nordisk veterinaermedicin. 31:97 105. Dohoo IR, Martin S. 1984. Subclinical ketosis: prevalence and associations with production and disease. Can J Comparative Med. 48(1):1 5. Dohoo IR, Martin SW, McMillan I, Kennedy BW. 1984. Disease, production and culling in Holstein-Friesian cows II. Age, season and sire effects. Preventive Veterinary Med. 2:655 670. Drackley JK, Veenhuizen JJ, Richard MJ, Young JW. 1991. Metabolic changes in blood and liver of dairy cows during either feed restriction or administration of 1,3-butanediol. J Dairy Sci. 74(12):4254 4264. Duffield T. 2000. Subclinical ketosis in lactating dairy cattle. Veterinary Clin North Am Food Animal Pract. 16:231 253. Gardner IA, Greiner M. 2006. Receiver operating characteristic curves and likelihood ratios: improvements over traditional methods for the evaluation and application of veterinary clinical pathology tests. Veterinary Clin Pathol. 35:8 17. Gathercole LL, Morgan SA, Tomlinson JW. 2012. Hormonal regulation of lipogenesis. Vitamins Hormones. 91:1 27. Giannini E, Botta F, Fasoli A, Ceppa P, Risso D, Lantieri PB, Celle G, Testa R. 1999. Progressive liver functional impairment is associated with an increase in AST/ALT ratio. Digestive Diseases Sci. 44:1249 1253. Gonzalez FD, Mui~no R, Pereira V, Campos R, Benedito JL. 2011. Relationship among blood indicators of lipomobilization and hepatic function during early lactation in highyielding dairy cows. J Veterinary Sci. 12:251 255. Ingvartsen KL. 2006. Feeding-and management-related diseases in the transition cow: Physiological adaptations around
calving and strategies to reduce feeding-related diseases. Animal Feed Sci Technol. 126:175 213. Katoh N. 2002. Relevance of apolipoproteins in the development of fatty liver and fatty liver-related peripartum diseases in dairy cows. J Veterinary Med Sci. 64(4):293 307. Kew MC. 2000. Serum aminotransferase concentration as evidence of hepatocellular damage. Lancet. 355:591 592. Kronfeld DS. 1965. Plasma non-esterified fatty acid concentrations in the dairy cow: responses to nutritional and hormonal stimuli, and significance in ketosis. Veterinary Rec. 9 (77):30 35. LeBlanc S. 2010. Monitoring metabolic health of dairy cattle in the transition period. J Reprod Dev. 56:S29 S35. Markusfeld O. 1985. Relationship between overfeeding, metritis and ketosis in high yielding dairy cows. Veterinary Rec. 116 (18):489 491. Moallem U, Bruckental I, Sklan D. 2004. Effect of feeding pregnant and non-lactating dairy cows a supplement containing a high proportion of non-structural carbohydrates on post-partum production and peripartum blood metabolites. Animal Feed Sci Technol. 116:185 195. Mulligan F, O’Grady L, Rice D, Doherty M. 2006. A herd health approach to dairy cow nutrition and production diseases of the transition cow. Animal Reprod Sci. 96:331 353. Oetzel GR. 2004. Monitoring and testing dairy herds for metabolic disease. Veterinary Clin North Am. 20:651 674. Ospina P, Nydam D, Stokol T, Overton T. 2010. Evaluation of nonesterified fatty acids and b-hydroxybutyrate in transition dairy cattle in the northeastern United States: Critical thresholds for prediction of clinical diseases. J Dairy Sci. 93:546 554. Patel T, Gores GJ. 1995. Apoptosis and hepatobiliary disease. Hepatology. 21:1725 1741. Rukkwamsuk T, Wensing T, Kruip TA. 1999. Relationship between triacylglycerol concentration in the liver and first ovulation in postpartum dairy cows. Theriogenology. 51:1133 1142. Sheth SG, Flamm SL, Gordon FD, Chopra S. 1998. AST/ALT ratio predicts cirrhosis in patients with chronic hepatitis C virus infection. Am J Gastroenterol. 93:44 48. Steen A, Gr€ onst€ ol H, Torjesen P. 1997. Glucose and insulin responses to glucagon injection in dairy cows with ketotis and fatty live. J Veterinary Med. 44:521 530. Watson ED, Williams LA. 1987. Influence of liver fat on postpartum hormone profiles on dairy cows. Animal Prod. 45:9 14. Xia C, Wang Z, Liu G, Zhang H, Xu C. 2010. Changes of plasma metabolites, hormones, and mRNA expression of liver PEPCK-C in spontaneously ketotic dairy cows. AsianAustralas J Animal Sci. 23(1):47 51
Veterinary Quarterly Publication details, including instructions for authors and subscription information: http://www.tandfonline.com/loi/tveq20
Critical thresholds of liver function parameters for ketosis prediction in dairy cows using receiver operating characteristic (ROC) analysis a
a
a
a
a
a
ab
Yuhang Sun , Bo Wang , Shi Shu , Hongyou Zhang , Chuang Xu , Ling Wu & Cheng Xia a
Department of College of Animal Science and Veterinary Medicine, Heilongjiang BaYi Agriculture University, Daqing 163319, China b
Department of Synergetic Innovation Center of Food Safety and Nutrition, Northeast Agricultural University, Harbin 150030, China Published online: 02 Apr 2015.
Click for updates To cite this article: Yuhang Sun, Bo Wang, Shi Shu, Hongyou Zhang, Chuang Xu, Ling Wu & Cheng Xia (2015): Critical thresholds of liver function parameters for ketosis prediction in dairy cows using receiver operating characteristic (ROC) analysis, Veterinary Quarterly, DOI: 10.1080/01652176.2015.1028657 To link to this article: http://dx.doi.org/10.1080/01652176.2015.1028657
PLEASE SCROLL DOWN FOR ARTICLE Taylor & Francis makes every effort to ensure the accuracy of all the information (the “Content”) contained in the publications on our platform. However, Taylor & Francis, our agents, and our licensors make no representations or warranties whatsoever as to the accuracy, completeness, or suitability for any purpose of the Content. Any opinions and views expressed in this publication are the opinions and views of the authors, and are not the views of or endorsed by Taylor & Francis. The accuracy of the Content should not be relied upon and should be independently verified with primary sources of information. Taylor and Francis shall not be liable for any losses, actions, claims, proceedings, demands, costs, expenses, damages, and other liabilities whatsoever or howsoever caused arising directly or indirectly in connection with, in relation to or arising out of the use of the Content. This article may be used for research, teaching, and private study purposes. Any substantial or systematic reproduction, redistribution, reselling, loan, sub-licensing, systematic supply, or distribution in any form to anyone is expressly forbidden. Terms & Conditions of access and use can be found at http:// www.tandfonline.com/page/terms-and-conditions
Veterinary Quarterly, 2015 http://dx.doi.org/10.1080/01652176.2015.1028657
SHORT COMMUNICATION Critical thresholds of liver function parameters for ketosis prediction in dairy cows using receiver operating characteristic (ROC) analysis Yuhang Suna, Bo Wanga,†, Shi Shua, Hongyou Zhanga, Chuang Xua, Ling Wua and Cheng Xiaa,b* a
Department of College of Animal Science and Veterinary Medicine, Heilongjiang BaYi Agriculture University, Daqing 163319, China; b Department of Synergetic Innovation Center of Food Safety and Nutrition, Northeast Agricultural University, Harbin 150030, China
Downloaded by [New York University] at 03:58 09 June 2015
(Received 28 May 2014; accepted 9 March 2015) Background: Fatty liver syndrome and ketosis are important metabolic disorders in high-producing cows during early lactation with fatty liver usually preceding ketosis. To date, parameters for early prediction of the risk of ketosis have not been investigated in China. Objective: To determine the predictive value of some parameters on the risk of ketosis in China. Animals and methods: In a descriptive study, 48 control and 32 ketotic Holstein Friesian cows were randomly selected from one farm with a serum b-hydroxybutyrate (BHBA) concentration of 1.20 mmol/L as cutoff point. The risk prediction thresholds for ketosis were determined by receiver operating characteristic (ROC) analysis. Results: In line with a high BHBA concentration, blood glucose concentration was significantly lower in ketotic cows compared to control animals (2.77 § 0.24 versus 3.34 § 0.03 mmol/L; P D 0.02). Thresholds were more than 0.76 mmol/L for nonesterified fatty acids (NEFA, with 65% sensitivity and 92% specificity), more than 104 U/L for aspartate aminotransferase (AST, 74% and 85%, respectively), less than 140 U/L for cholinesterase (CHE, 75% and 59%, respectively), and more than 3.3 mmol/L for total bilirubin (TBIL, 58% and 83%, respectively). There were significant correlations between BHBA and glucose (R D 4.74), or CHE (R D 0.262), BHBA and NEFA (R D 0.520), or AST (R D 0.525), or TBIL (R D 0.278), or direct bilirubin (DBIL, R D 0.348). Conclusions: AST, CHE, TBIL and NEFA may be useful parameters for risk prediction of ketosis. Clinical importance: This study might be of value in addressing novel directions for future research on the connection between ketosis and liver dysfunction. Keywords: bovine; dairy cows; ketosis; liver function; risk prediction; threshold; ROC analysis
1. Introduction Fatty liver syndrome and ketosis are some important metabolic disorders in high-producing dairy cows especially during early lactation. When energy intake cannot meet the energy demand of the mammary gland for milk production and other metabolic functions, fat mobilization from adipose tissue rapidly increases. This releases a large amount of nonesterified fatty acids (NEFA) into the blood circulation, which are transported to the liver. Given that the oxidizing ability of the liver regarding NEFA in dairy cows is limited, excessive NEFA in the liver are converted into ketone bodies, or accumulated in hepatocytes in the form of triacylglycerol leading to ketosis and/or fatty liver syndrome (Rukkwamsuk et al. 1999, Gathercole et al. 2012). As a consequence, fatty liver syndrome is closely related to ketosis. Usually, the fatty liver syndrome is also associated with abnormalities in liver function, which result in reduction of NEFA oxidization in the tricarboxylic acid cycle and the synthesis of lipoprotein, and as a consequence the blood concentrations of ketone bodies increase rapidly and persistently. Ketosis is experimentally induced by fasting (Brumby et al. 1975) and by administration of 1,3-butanediol (Veenhuizen & co-workers 1991). In both cases, fatty
liver precedes ketosis. On the other hand, fatty liver is found in almost all cases of natural ketosis. As in fatty liver syndrome, overfeeding is epidemiologically suspected as a major causal factor for ketosis (Markusfeld 1985). However, it should be realized that in normal cows, triacylglycerol accumulation in the liver at the end of the dry period is physiological. A large concentration of ketone bodies in the blood may cause anorexia, or decreased feed intake, which worsens the negative energy balance further, to provoke severe ketosis and fatty liver. Affected cows are unhealthy and show poor production performance, including decreases in dry matter intake (DMI) and milk yield (MY) (Dohoo & Martin 1984, Dohoo et al. 1984), which cause financial losses and poor animal welfare. The concentrations of bile components (e.g. bilirubin, bile acids and cholesterylesters) and triglyceride (TG) increase in the plasma of dairy cows with fatty liver syndrome, while the concentrations of free cholesterol, citrate and glycogen decrease. Fatty liver syndrome is usually characteristic of cows with high blood concentrations of NEFA, butyrate and acetoacetic acid, which may be toxic to cells and reduce b-oxidation and gluconeogenesis in the liver (Ametaj et al. 2005). In addition to the increase
*Corresponding author. Email: [email protected] † Parallel first author: Bo Wang This article was originally published with errors. This version has been corrected. Please see Erratum (http://dx.doi.org/10.1080/ 01652176.2015.1040689). Ó 2015 Taylor & Francis
Downloaded by [New York University] at 03:58 09 June 2015
2
Y. Sun et al.
in blood NEFA concentrations, decreased concentrations in cholesterylesters and phospholipids are observed in cows with ketosis (Kronfeld 1965, Watson & Williams 1987, Katoh 2002). However, current detection of fatty liver is possible only by minor surgery and subsequent biopsy analysis (Bobe & co-workers 2005). Some blood parameters, such as total protein (TP), albumin (ALB), cholinesterase (CHE) and prothrombin time, are commonly considered to reflect the synthetic ability of the liver because they are only synthesized in hepatocytes and have a positive association with their synthetic ability. Once liver synthetic capability decreases, their values in plasma also decease. In addition, the activities of aspartate aminotransferase (AST or GOT), alanine aminotransferase (ALT or GPT), adenosine deaminase and lactic dehydrogenase (LDH) in plasma are often used as biomarkers that reflect whether hepatocytes are damaged and the degree of hepatocyte damage. When liver cell membranes are impaired or liver cell necrosis occurs, these enzymes are released into the blood. Therefore, assay of enzyme activity in plasma or serum can predict the degree of exposure in hepatic cells (Steen et al. 1997). ALT is the most sensitive target in the detection of acute liver damage and AST is a more sensitive indicator of the degree of exacerbated liver damage. When 1% of liver cells have undergone necrosis, the serum ALT activity doubles. However, the activity of AST increases continuously to exceed the value of ALT; this suggests that severe damage has occurred to liver parenchyma and is an indication of exacerbated illness (Giannini et al. 1999). Given that abnormalities of hepatic function may be observed in relation to excretion, secretion, transportation and detoxication, the plasma concentrations of total bilirubin (TBIL), direct bilirubin (DBIL), total bile acid and blood ammonia usually increase (Patel & Gores 1995). Among the available parameters, it is necessary to select a sensitive, specific, low-cost and simple target to assess liver function. It is accepted that the blood concentration of b-Hydroxybutyrate (BHBA) is the gold standard for the diagnosis of ketosis because blood BHBA has high sensitivity, specificity, accuracy and stability (Duffield 2000, Atkinson et al. 2004). A serum BHBA concentration above 1.2 mmol/L is commonly used as a cutoff point for detection of ketosis (LeBlanc 2010, Ospina et al. 2010). Previous reports on the critical thresholds for ketosis prediction have incorporated receiver operating characteristic (ROC) analysis (Ospina et al. 2010, Asl et al. 2011). Mulligan et al. suggested that milk data may be used to assess the energy balance in early lactation; other assessments include feed intake and refusal of intake, blood BHBA and NEFA concentrations 14 days prepartum, and on-farm approaches to control postpartum ketosis (Mulligan et al. 2006). Although liver dysfunction is closely associated with fatty liver and ketosis, it is unclear whether liver function parameters can predict ketosis. To date, the cutoff point of liver function parameters for early prediction of the risk of ketosis has not been investigated in China. Therefore, the objective of this study was to investigate the
relationship between liver function and ketosis in China using binary logistic regression analysis, and to determine the critical thresholds of liver function parameters for ketosis prediction using ROC curves in order to provide a scientific basis for the prevention of ketosis in dairy cows.
2. Materials and methods 2.1. Animals The study was carried out in a dairy herd comprising more than 3000 Holstein Friesian cows in total that was maintained in a free-stall housing using a complete dairy management software (Valley Agricultural Software, Tulare, CA, USA) system in Heilongjiang, China. The average calving interval of the herd was 444 days and the mean (305 days) MY 23.6 kg/day associated with 4.55% fat and 3.30% protein. From April to June 2012, 80 multiparous healthy cows and cows with ketosis were selected randomly generating 48 control cows and 32 ketotic cows; basic information and MY were recorded. The cows were considered to have ketosis (K) if they had serum BHBA concentrations >1.20 mmol/L, otherwise they were regarded as healthy controls (C) (Oetzel 2004). All the cows were fed a total mixed ration during the transition period, which consisted of 8.5 kg concentrated feed, 18.5 kg maize silage, 4 kg hay and 0.35 kg fat. Its nutritional levels were 57% dry matter (DM), 17% crude protein, 1.76 mcal/DM net energy for lactation, 5.6% fat, 38.5% neutral detergent fiber, 20.3% acid detergent fiber, 180 g Ca, and 100 g P, with a daily DMI of 19 kg. 2.2.
Sample collection
Blood samples were taken from the caudal vein into a plain tube and a tube containing EDTA (ethylene diamine tetraacetic acid) between 14 and 21 days postpartum, before the morning feed. EDTA blood samples were immediately centrifuged at 1400 g for 5 min at room temperature. The supernatants were aliquoted into Eppendorf tubes (1 ml plasma/tube) and stored at 80 C until analysis. All haemolytic samples were discarded. This study was carried out in strict accordance with the recommendations in the Guide for the Care and Use of Laboratory Animals of the National Institutes of Health, USA. All animal experiments were conducted according to the practices and standards approved by the Animal Welfare and Research Ethics Committee at Heilongjiang Bayi Agricultural University, China. 2.3. Laboratory analysis BHBA was measured by means of an enzyme-linked immunosorbent assay (Nanjing Senbeijia Biotechnology Company, Nanjing China); AST, CHE and ALT were detected by the velocity method (Beijing Strong Biotechnology Company, Beijing, China). NEFA were measured by an enzymic method (Nanjing Senbeijia Biotechnology Company, Nanjing China); glucose (Glc) was detected by
Veterinary Quarterly the glucose oxidase method; TBIL and DBIL were analyzed by the dyeline method (Shanghai Desai Biotechnology Company, Shanghai, China). TP and ALB were measured by a colorimetric method (ACCU-CHEK, Roche, Basel, Switzerland). All the plasma samples were analyzed using an automatic biochemical analyzer (HITACHI 7600, Kyoto, Japan).
Downloaded by [New York University] at 03:58 09 June 2015
2.4.
Statistical analysis
Significant analysis of several parameters: student’s t-test was used to analyze any significant differences in plasma concentrations of Glc, BHBA, NEFA, ALT, AST, CHE, TBIL, DBIL, indirect bilirubin (IBIL), TP, ALB, globulin (GLB) and MY between the healthy group and the ketotic group. The analyses were performed using SPSS software (Version 19.0, IBM, New York, USA). All data were presented as mean and standard error (SE) in Table 1, and P < 0.05 was considered to represent statistical significance. Pearson’s correlation coefficients: given that blood BHBA concentration has been accepted as gold standard for the diagnosis of ketosis and was also considered as an optimal herd monitoring biomarker (Oetzel 2004), Pearson’s correlation coefficients were utilized to reveal the correlations between BHBA and the parameters in Table 1. Regression analysis: subsequently, two logistic models were established to screen risk factors for negative energy balance associated with ketosis. Model 1 included Glc and NEFA to determine the causal relationship between energy balance and ketosis, and other risk factors for liver dysfunction included in Model 2 were AST and CHE, to confirm the causal relationship between liver function and ketosis.
Table 1. Mean (§ standard error) and P-value for milk yield and 12 blood biochemical parameters in the control and ketosis groups. Parameter Number MY (kg/day) BHBA (mmol/L) NEFAs (mmol/L) Glc (mmol/L) ALT (U/L) AST (U/L) CHE (U/L) TP (g/L) ALB (g/L) GLB (g/L) TBIL (mmol/L) DBIL (mmol/L) IBIL (mmol/L)
C
K
P-value
48 33.42 § 1.02 0.61 § 0.02 0.54 § 0.04 3.34 § 0.03 21.94 § 0.74 89.27 § 2.82 153.69 § 2.95 78.69 § 0.92 34.12 § 0.36 44.51 § 1.13 2.75 § 0.11 1.79 § 0.09 0.95 § 0.07
32 33.05 § 1.96 2.84 § 0.21 0.87 § 0.06 2.77 § 0.24 22.06 § 0.89 134.84 § 8.63 139.03 § 3.58 75.46 § 0.88 32.78 § 0.51 42.71 § 0.87 3.61 § 0.28 2.51 § 0.18 1.09 § 0.13
0.87 0.00 0.00 0.02 0.91 0.00 0.00 0.02 0.03 0.21 0.01 0.00 0.31
Note: K: ketosis group, C: control group; MY: milk yield, Glc: glucose, BHBA: b-hydroxybutyrate, NEFAs: nonesterified fatty acids, ALT: alanine aminotransferase, AST: aspartate aminotransferase, CHE: cholinesterase, TP: total protein, ALB: albumin, GLB: globulin, TBIL: total bilirubin, DBIL: direct bilirubin, IBIL: indirect bilirubin.
3
ROC analysis: finally, the cutoff points or critical thresholds of the risk factors were determined using ROC analysis, in which BHBA was included as a continuous outcome parameter. Furthermore, curves of sensitivity versus specificity were plotted using ROC analysis for NEFA, AST, CHE, DBIL, etc., to include all possible threshold values of the parameters assessed. Youden’s index was computed to determine the optimal thresholds of the selected parameters. The area under the curve (AUC) was used to assess the diagnostic precision of the eligible parameters (Gardner & Greiner 2006).
3. Results 3.1. Significance test The clinical data and biochemical parameters are presented in Table 1. The analysis included both subclinical cases and clinical cases, which may have caused the wide range observed in MY and the median values of the two groups to be similar. A box plot (Figure 1) is presented to demonstrate this phenomenon. 3.2.
Pearson’s correlation coefficients
In the 80 cows tested, there was a significant negative correlation between the plasma concentrations of BHBA and Glc (R D 4.74, P D 0.000), or CHE (R D 0.262, P D 0.019), and also a prominent positive correlation between the plasma concentrations of BHBA and NEFA (R D 0.520, P D 0.000), AST (R D 0.525, P D 0.000), TBIL (R D 0.278, P D 0.012) or DBIL (R D 0.348, P D 0.002). However, there was no significant correlation between the plasma concentrations of BHBA and ALT, TP, ALB, GLB or IBIL (P 0.05). 3.3.
Regression analysis
In the first logistic model, plasma Glc (P D 0.101) was removed and NEFA (P D 0.000) was retained, suggesting that an increase in NEFA can predict the rise in BHBA. In the second logistic model, plasma AST (P D 0.000), CHE
Figure 1. Box plot of milk yield in Holstein Friesian cows for control group (n D 48) and ketosis group (n D 32).
4
Y. Sun et al.
Table 2. Prediction point, sensitivity, specificity, respective SE, CLR and AUC for NEFAs, AST, CHE and TBIL (n D 48 control and 32 ketotic cows). Parameter NEFAs (mmol/L) AST (U/L) CHE (U/L) TBIL (mmol/L)
Prediction point
Sensitivity (%)
SE (sen.%)
Specificity (%)
SE (spe.%)
CLR
AUC
0.76 104.0 139.5 3.25
64.5 74.2 75.0 58.1
9.88 8.13 8.82 9.73
91.7 85.4 59.4 83.3
3.69 5.04 6.25 5.07
7.77 5.08 1.85 3.48
0.818, p D 0.000 0.849, p D 0.000 0.704, p D 0.002 0.714, p D 0.001
Note: SE: standard error, CLR: positive likelihood ratio; NEFAs: nonesterified fatty acids, AST: aspartate aminotransferase, CHE: cholinesterase, TBIL: total bilirubin, AUC: area under ROC curve.
Downloaded by [New York University] at 03:58 09 June 2015
(P D 0.001) and TBIL (P D 0.034) were retained and DBIL (P D 0.354) was eliminated. 3.4. ROC analysis The prediction thresholds, sensitivity, specificity, respective SE, positive likelihood ratios (CLR) and areas under the ROC curves for NEFA, AST, CHE and TBIL are presented in Table 2. The optimal prediction thresholds were determined by Youden’s index to be more than 0.76 mmol/L for NEFA, with 65% sensitivity and 92% specificity, more than 104 U/L for AST, with 74% sensitivity and 85% specificity, less than 140 U/L for CHE, with 75% sensitivity and 59% specificity, and more than 3.25 mmol/L for TBIL, with 58% sensitivity and 83% specificity. The area under the ROC curves for NEFA, AST, CHE and TBIL are shown in Figure 2. As CHE had a negative correlation with BHBA, the area above the ROC curve was selected. The area under (or above) the ROC curve was greater than 0.70 for all four parameters (Table 2). 4. Discussion As shown in Table 1, when cows with ketosis were compared with healthy controls, there were significant increases in mean plasma concentrations of BHBA, NEFA, AST, TBIL and DBIL (P < 0.01), obvious decreases in plasma CHE (P < 0.01), Glc, TP and ALB
Figure 2. Receiver operating characteristic (ROC) plots for NEFA, AST, CHE and TBIL for diagnosis of ketosis in Holstein Friesian cows (NEFAs: nonesterified fatty acids; AST: aspartate aminotransferase; CHE: cholinesterase; TBIL: total bilirubin).
(P < 0.05), but no significant change in plasma concentrations of IBIL, GLB, ALT and MY. In addition, in subsequent Pearson’s correlation coefficients analysis, there was a significant correlation between the plasma concentrations of BHBA and glucose, CHE, NEFA, AST, TBIL or DBIL. Blood ketone body and glucose concentrations are the main diagnostic parameters for ketosis. An increased blood ketone body concentration accompanied by a decreased blood glucose concentration, becomes the main risk factor for ketosis (Ingvartsen 2006) and the low glucose concentration reflects the negative energy balance and triggers lipolysis due to the associated low insulin concentration. BHBA is the primary component of blood ketone bodies in dairy cattle; it is relatively stable in whole blood, serum or plasma, both in vitro and in vivo (Custer et al. 1983). Thus, blood BHBA concentration has been accepted as gold standard for the diagnosis of ketosis and was also considered as an optimal herd monitoring biomarker (Oetzel 2004). As a result, in this study BHBA, glucose, NEFA, AST, TBIL, DBIL and CHE were considered to be associated with ketosis, and can be regarded as novel biomarkers for diagnosis of ketosis. However, it should be realized that blood concentrations of BHBA and NEFA in early lactation are higher than in mid-lactation. The prevalence of a high level of NEFA in early lactation has been reported to be 67%, but was only 17% in mid-lactation, whereas the prevalence of subclinical ketosis in early lactation may reach 41%, but in mid-lactation 28% (Gonzalez et al. 2011). In the subsequent regression analysis, NEFA was retained and glucose was removed. This is in accordance with the finding of Asl that NEFA has a predictive effect for ketosis but glucose does not (Asl et al. 2011). In the second logistic model, several parameters associated with liver function, namely AST, CHE and TBIL were retained. It implied that the plasma levels of NEFA, AST, CHE and TBIL may predict the risk of ketosis, and liver dysfunction may become an important risk factor for ketosis. In the first months of lactation, the plasma concentrations of NEFA and TBIL should decrease gradually, while total lipids should increase gradually in healthy dairy cows (Dale et al. 1979). In this study, the mean NEFA concentration was significantly different (P D 0.000) between the control and ketotic cows (0.54 and 0.87 mmol/L, respectively). The optimal threshold for NEFA was determined to be 0.76 mmol/L using ROC analysis, which is similar to the NEFA value of more than
Downloaded by [New York University] at 03:58 09 June 2015
Veterinary Quarterly 0.72 mmol/L reported by Ospina (Ospina et al. 2010). However, this result was higher than those of other experiments, such as the 0.4 mmol/L reported by Oetzel (Oetzel 2004) and 0.26 mmol/L by Asl (Asl et al. 2011). This phenomenon may be related to the high level of milk production (33 kg/day) and the high incidence of ketosis (40%) in the dairy herd studied herein. This suggests a need to adopt nutritional management and preventive measures to reduce the likelihood of ketosis. Plasma levels of ALT, AST and CHE, which are released from liver cells, can be used as important parameters that reflect whether hepatocyte damage has occurred, and the extent of such damage. When liver cell membranes are impaired or liver cell necrosis occurs, most of these enzymes are released into the blood. Assay of enzyme activity in plasma or serum can demonstrate the degree of damage to hepatic cells (Steen et al. 1997). Among clinical enzyme tests, AST and ALT may be used to investigate the type and degree of damage to liver cells. As mentioned before, ALT is the most sensitive target in the diagnosis of acute liver damage, whereas AST is more sensitive in reflecting the degree of damage (Kew 2000). The normal activity of AST in blood is 78 132 U/L in healthy cattle (Ingvartsen 2006). In this study, the mean activity of AST was 89 U/L in the control group. This result was close to that reported by Moallem (Moallem et al. 2004). AST exists primarily in mitochondria. When mild liver damage occurs, the increase of AST is lower than that of ALT, i.e. AST/ALT 0.05). So, the correlation between TP, ALB or GLB and liver function is not substantiated yet. In this study, blood TBIL, DBIL and IBIL were considered as indicators reflecting the capacity of the liver for excretion, secretion and detoxication. When hepatocytes are damaged, the hepatic capability for excretion, secretion and detoxication are reduced, and this results in the increase of the three parameters mentioned above. The TBIL (P D 0.007) and DBIL (P D 0.000) in the ketosis group were significantly higher than in the control group, but IBIL showed no difference between both groups. In the logistic regression analysis, only TBIL (P D 0.034) stayed in the model; DBIL (P D 0.703) did not have any predictive value for ketosis. The prediction threshold of TBIL was more than 3.3 mmol/L, with 58% sensitivity and 83% specificity. Therefore, a plasma concentration of TBIL higher than 3.3 mmol/L suggests abnormal liver function, and can also be considered as a predictor for a high risk of ketosis. In conclusion, the results of this study showed that the prevalence of ketosis, including both clinical and subclinical ketosis, in dairy cattle on a farm in Heilongjiang Province, China, is significant. For their greater sensitivity and specificity, the values of NEFA, AST, CHE and TBIL can be used during early lactation for prediction of the risk of ketosis and liver dysfunction, which will not only provide new references for prediction and diagnosis of ketosis, but also finds novel directions for research of the connection between ketosis and liver dysfunction. Furthermore, to prevent the economic loss due to ketosis associated with liver dysfunction, early treatment of affected cows is necessary. Prediction and prevention of ketosis must be achieved through monitoring of liver function and proper nutritional programs for dry cows and those in early lactation. Disclosure statement No potential conflict of interest was reported by the authors.
Funding The authors thank the National Science Foundation Committee [31001094]; the China and National Science and Technology Project ‘Twelfth Five-Year’ in rural areas [2012BAD12B05-2], [2012BAD12B03-2]; the National Science and Technology Support Program [2013BAD21B01]; the National Science Foundation Committee [31672625]; China Spark Program [2012GA670001] and NCET [1252-NCET-003] for the financial support.
6
Y. Sun et al.
Downloaded by [New York University] at 03:58 09 June 2015
References Ametaj B, Bradford B, Bobe G, Nafikov R, Lu Y, Young J, Beitz D. 2005. Strong relationships between mediators of the acute phase response and fatty liver in dairy cows. Can J Animal Sci. 85:165 175. Asl AN, Nazifi S, Ghasrodashti AR, Olyaee A. 2011. Prevalence of subclinical ketosis in dairy cattle in the Southwestern Iran and detection of cutoff point for NEFA and glucose concentrations for diagnosis of subclinical ketosis. Preventive Veterinary Med. 100:38 43. Atkinson SJ, Hosford MA, Molitoris BA. 2004. Mechanism of actin polymerization in cellular ATP depletion. J Biol Chem. 279:5194 5199. Brumby PE, Anderson M, Tuckley B, Storry JE, Hibbit KG. 1975. Lipid metabolism in the cow during starvationinduced ketosis. Biochem J. 146(3):609 615. Custer E, Myers J, Poffenbarger P, Schoen I. 1983. The storage stability of 3-hydroxybutyrate in serum, plasma, and whole blood. Am J Clin Pathol. 80:375 380. Dale H, Vik-Mo L, Fjellheim P. 1979. A field survey of fat mobilization and liver function of dairy cows during early lactation. Nordisk veterinaermedicin. 31:97 105. Dohoo IR, Martin S. 1984. Subclinical ketosis: prevalence and associations with production and disease. Can J Comparative Med. 48(1):1 5. Dohoo IR, Martin SW, McMillan I, Kennedy BW. 1984. Disease, production and culling in Holstein-Friesian cows II. Age, season and sire effects. Preventive Veterinary Med. 2:655 670. Drackley JK, Veenhuizen JJ, Richard MJ, Young JW. 1991. Metabolic changes in blood and liver of dairy cows during either feed restriction or administration of 1,3-butanediol. J Dairy Sci. 74(12):4254 4264. Duffield T. 2000. Subclinical ketosis in lactating dairy cattle. Veterinary Clin North Am Food Animal Pract. 16:231 253. Gardner IA, Greiner M. 2006. Receiver operating characteristic curves and likelihood ratios: improvements over traditional methods for the evaluation and application of veterinary clinical pathology tests. Veterinary Clin Pathol. 35:8 17. Gathercole LL, Morgan SA, Tomlinson JW. 2012. Hormonal regulation of lipogenesis. Vitamins Hormones. 91:1 27. Giannini E, Botta F, Fasoli A, Ceppa P, Risso D, Lantieri PB, Celle G, Testa R. 1999. Progressive liver functional impairment is associated with an increase in AST/ALT ratio. Digestive Diseases Sci. 44:1249 1253. Gonzalez FD, Mui~no R, Pereira V, Campos R, Benedito JL. 2011. Relationship among blood indicators of lipomobilization and hepatic function during early lactation in highyielding dairy cows. J Veterinary Sci. 12:251 255. Ingvartsen KL. 2006. Feeding-and management-related diseases in the transition cow: Physiological adaptations around
calving and strategies to reduce feeding-related diseases. Animal Feed Sci Technol. 126:175 213. Katoh N. 2002. Relevance of apolipoproteins in the development of fatty liver and fatty liver-related peripartum diseases in dairy cows. J Veterinary Med Sci. 64(4):293 307. Kew MC. 2000. Serum aminotransferase concentration as evidence of hepatocellular damage. Lancet. 355:591 592. Kronfeld DS. 1965. Plasma non-esterified fatty acid concentrations in the dairy cow: responses to nutritional and hormonal stimuli, and significance in ketosis. Veterinary Rec. 9 (77):30 35. LeBlanc S. 2010. Monitoring metabolic health of dairy cattle in the transition period. J Reprod Dev. 56:S29 S35. Markusfeld O. 1985. Relationship between overfeeding, metritis and ketosis in high yielding dairy cows. Veterinary Rec. 116 (18):489 491. Moallem U, Bruckental I, Sklan D. 2004. Effect of feeding pregnant and non-lactating dairy cows a supplement containing a high proportion of non-structural carbohydrates on post-partum production and peripartum blood metabolites. Animal Feed Sci Technol. 116:185 195. Mulligan F, O’Grady L, Rice D, Doherty M. 2006. A herd health approach to dairy cow nutrition and production diseases of the transition cow. Animal Reprod Sci. 96:331 353. Oetzel GR. 2004. Monitoring and testing dairy herds for metabolic disease. Veterinary Clin North Am. 20:651 674. Ospina P, Nydam D, Stokol T, Overton T. 2010. Evaluation of nonesterified fatty acids and b-hydroxybutyrate in transition dairy cattle in the northeastern United States: Critical thresholds for prediction of clinical diseases. J Dairy Sci. 93:546 554. Patel T, Gores GJ. 1995. Apoptosis and hepatobiliary disease. Hepatology. 21:1725 1741. Rukkwamsuk T, Wensing T, Kruip TA. 1999. Relationship between triacylglycerol concentration in the liver and first ovulation in postpartum dairy cows. Theriogenology. 51:1133 1142. Sheth SG, Flamm SL, Gordon FD, Chopra S. 1998. AST/ALT ratio predicts cirrhosis in patients with chronic hepatitis C virus infection. Am J Gastroenterol. 93:44 48. Steen A, Gr€ onst€ ol H, Torjesen P. 1997. Glucose and insulin responses to glucagon injection in dairy cows with ketotis and fatty live. J Veterinary Med. 44:521 530. Watson ED, Williams LA. 1987. Influence of liver fat on postpartum hormone profiles on dairy cows. Animal Prod. 45:9 14. Xia C, Wang Z, Liu G, Zhang H, Xu C. 2010. Changes of plasma metabolites, hormones, and mRNA expression of liver PEPCK-C in spontaneously ketotic dairy cows. AsianAustralas J Animal Sci. 23(1):47 51
