This work was performed at the Clinic for Cattle, University of Veterinary Medicine Hannover, Foundation
Effects of Dexamethasone-21-Isonicotinate on Peripheral Insulin Action in Dairy Cows 5 days after Surgical Correction of Abomasal Displacement
Version of Record online: 29 NOV 2012
Copyright © 2012 by the American College of Veterinary Internal Medicine
Journal of Veterinary Internal Medicine
Volume 27, Issue 1, pages 200–206, January/February 2013
How to Cite
Kusenda, M., Kaske, M., Piechotta, M., Locher, L., Starke, A., Huber, K. and Rehage, J. (2013), Effects of Dexamethasone-21-Isonicotinate on Peripheral Insulin Action in Dairy Cows 5 days after Surgical Correction of Abomasal Displacement. Journal of Veterinary Internal Medicine, 27: 200–206. doi: 10.1111/jvim.12010
- Issue online: 11 JAN 2013
- Version of Record online: 29 NOV 2012
- Manuscript Accepted: 26 SEP 2012
- Manuscript Revised: 30 JUN 2012
- Manuscript Received: 31 DEC 2011
- Hyperinsulinemic-euglycemic clamp;
- Insulin sensitivity;
Dexamethasone frequently is used for treatment of ketosis in dairy cows, but its effects are not fully understood.
Dexamethasone treatment affects whole body insulin sensitivity.
Twelve German Holstein cows, 2–4 weeks postpartum, 5 days after omentopexy to correct left abomasal displacement.
Randomized, blinded, case-control study. Treatment with dexamethasone-21-isonicotinate (DG; 40 μg/kg IM; n = 6) or saline (control group [CG], 15 mL IM, n = 6) on day 0 (d0). Blood samples were obtained before (d0) and after treatment (d1 and d2), and analyzed for glucose, insulin, and nonesterified fatty acid (NEFA) concentrations. Hepatic triglycerides (TAG) were measured in liver samples taken on d0 and d2. Five consecutive hyperinsulinemic-euglycemic clamps (HEC-I-V; insulin dosages: 0.1, 0.5, 2, 5, 10 mU/kg/min, respectively) were performed on d1 and steady state glucose infusion rate (SSGIR), insulin concentration (SSIC), insulin sensitivity index (ISI = SSGIR/SSIC), and plasma NEFA concentration (SSNEFA) were assessed.
Compared with CG-cows, DG-cows on d1 had higher plasma glucose (P = .004) and insulin (P < .001) concentrations, decreased SSGIR (HEC-II, P = .002; HEC-IV, P = .033), ISI (HEC-I, P < .015; HEC-II, P = .004), and insulin-stimulated decrease in SSNEFA (HEC-II, P = .006; HEC-III, P = .01; HEC-IV, P = .003; HEC-V, P = .011). Decrease in hepatic TAG content in DG-cows was higher compared with CG-cows (P < .001).
Conclusions and Clinical Importance
Dexamethasone decreases whole body insulin sensitivity and affects glucose and lipid metabolism in early lactating dairy cows.
adipose triglycerides lipase
body condition score
coefficient of variation
- d0, d1, d2
day 0, day 1, day 2
glucose transporter 4
hormone sensitive lipase
insulin sensitivity index
left displacement of the abomasum
nonesterified fatty acid
steady-state glucose infusion rate
steady-state plasma insulin concentration
steady-state plasma nonesterified fatty acid concentration
During early lactation, excessive lipomobilization with subsequent development of ketosis and fatty liver is a common metabolic disorder in high-yielding dairy cows. Glucocorticoids such as dexamethasone frequently are used in treatment protocols for ketosis. The efficacy of glucocorticoids has been clearly demonstrated in experimental[1, 2] and clinical field studies.
However, the underlying mechanisms of glucocorticoid effects in dairy cows are not fully understood. For example, it appears unlikely that glucocorticoid-induced hyperglycemia is caused by increased glucose production because key enzymes of hepatic gluconeogenesis are not stimulated by the glucocorticoid dexamethasone in neonatal calves.[4, 5] Furthermore, dexamethasone does not affect endogenous glucose production in calves and adult cattle. Induced hyperglycemia in dairy cows also could be caused by peripheral insulin resistance. The complex effects of glucocorticoids on insulin resistance with subsequent alterations in insulin-mediated whole-body glucose uptake and antilipolytic action of insulin have been reported in humans,[8-10] horses,[11-13] and calves.[6, 14] Although marked differences in digestion of carbohydrates and regulation of glucose homeostasis exist between monogastric animals and ruminating cattle, effects of dexamethasone on insulin resistance have not yet been studied in early lactating dairy cows.
The objective of this study was to evaluate the effects of dexamethasone on whole-body glucose consumption and the antilipolytic effect of insulin as indicators of insulin resistance by means of hyperinsulinemic-euglycemic clamp (HEC)[16-18] in dairy cows.
Materials and Methods
The study was performed at the Clinic for Cattle, University of Veterinary Medicine Hannover, and conducted according to the guidelines of the Research Animal Act (research permit number 33-42502-06/1084) of the Lower Saxony Federal State Office for Consumer Protection and Food Safety, Oldenburg, Germany.
Twelve multiparous German Holstein cows were included in the study. Animals had been admitted to the clinic for correction of left displacement of the abomasum (LDA) 2–4 weeks after calving. The cows were 3–8 years of age (5.0 ± 1.4 years), and weighted 430–684 kg (554 ± 71 kg). Body condition (BCS) was assessed on a 1–5 scale. Experiments were carried out 5 days after omentopexy for surgical correction of the LDA. Cows only were enrolled into the study if (a) general condition and feed intake after surgery were undisturbed, (b) daily milk yield was >15 kg (mean ± SD: 22.1 ± 5.5 kg), (c) no other diseases such as metritis, mastitis, or lameness were clinically detected, and (d) cows were neither treated with glucocorticoids within the last 10 days nor with glucose or glucose precursors within the last 24 hours. Cows were kept in individual pens on straw bedding and were fed corn silage, hay, and concentrates according to requirements. Water was available ad libitum. Cows were milked twice daily (6 am and 6 pm).
In a randomized, blinded, placebo-controlled study, each cow was investigated for 3 days (Table 1). On day 0 (d0) (5 days after LDA-surgery), liver biopsies were taken, both jugular veins were catheterized1 and blood samples (lithium-heparin) were obtained to assess pretreatment results. Subsequently, cows were randomly assigned to 1 of 2 groups and were treated with either 15-mL saline (control group CG; IM, n = 6) or dexamethasone2 (DG; 40 μg/kg dexamethasone-21-isonicotinate, IM, n = 6). On d1, 5 consecutive HECs were performed. On d2, a 2nd liver biopsy was obtained along with blood samples. All cows received antibiotics3 (30,000 IU/kg procaine penicillin G IM, q24h) from d0 to d2.
|0a||07:30||Blood sample (basal results)|
|12:00||Catheterization of both jugular veins|
|14:00||Treatmentb (CG; DG)|
|18:00||Withdrawal of concentrates and corn silage (hay and water still accessible)|
|1||07:30||3 blood samples in 10-minute intervals (clamp-baseline results)|
|08:00–10:00||HEC-I (0.1 mU/kg/min)|
|10:00–12:00||HEC-II (0.5 mU/kg/min)|
|12:00–14:00||HEC-III (2 mU/kg/min)|
|14:00–16:00||HEC-IV (5 mU/kg/min)|
|16:00–18:00||HEC-V (10 mU/kg/min)|
|18:00–22:00||Control period (glucose infusion, blood samples)|
|22:00||Infusion of saline, glucose, KCl for the next 8 hours|
Removal of catheters
The clamps were performed after withdrawal of silage and concentrates for 14 hours; water and hay were available to the animals ad libitum at any time.
Three blood samples were collected in 10 min intervals to assess clamp-basal concentrations of insulin, glucose, nonesterified fatty acids (NEFA), and beta-hydroxybutyrate (BHB). Thereafter, 5 consecutive HECs with increasing doses of insulin4 (HEC-I, II, III, IV, and V: 0.1, 0.5, 2, 5, and 10 mU/kg/min, respectively) were performed, each for 2 hours. Insulin solutions were infused into the right jugular vein with an infusion pump.5 Blood glucose concentration was determined every 10 minutes by analyzing a sample from the left jugular vein. Blood glucose was clamped at the clamp-baseline concentration by adjusted infusion of glucose solution6 via a 2nd pump into the right jugular vein. To avoid hypokalemia caused by hyperinsulinemia, potassium chloride7 was added to the insulin solutions infused during HEC-III (1 mmol KCl/min/cow) and HEC-IV as well as HEC-V (2 mmol KCl/min/cow). At 10, 20, 30, 60, 90, 100, 110, and 120 minutes after the start of each HEC, additional blood samples (lithium-heparin) were taken, stored on ice immediately and centrifuged within 1 hour (15 min, 1500 g, 4°C). The obtained plasma was stored at −80°C in Eppendorf cups until further analysis. After completion of HEC-V, blood samples were taken throughout the next 4-hour period (5, 10, 20, 30, 40, 60, 80, 120 180, and 240 minutes after the end of insulin infusion).
After the last collection of blood, all cows received an infusion overnight (10-L saline, 750-mL 40% glucose [wt/vol], 300-mL 1M KCl).
Blood and Tissue Analyses
For surveillance of euglycemia during the clamp, blood glucose concentration was assessed with glucometer.11 The coefficient of variation (CV) determined by analyzing 1 sample 12 times was 6.4%. For statistical evaluation, plasma concentrations of glucose, NEFA, and BHB were analyzed with an automatic analyzer12 with commercial enzymatic kits. Insulin concentration was measured by a commercial radioimmunoassay.13 The intra-assay and interassay CV were 5.8 and 14.2%, respectively. Triacylglyceride (TAG) and total lipid (TL) content in the hepatic tissue were assessed as described previously.
Calculations and Statistical Analyses
Steady-state glucose infusion rate (SSGIR) was assessed as mean infusion rate during the last 30 minutes of each 120-minute HEC. Accordingly, steady-state plasma insulin concentrations (SSIC) and steady-state NEFA concentrations (SSNEFA) were defined as mean plasma concentration of 3 blood samples taken within the last 30 minutes of each HEC. To demonstrate relative changes in plasma NEFA concentrations during the 5 HECs, NEFA concentrations also are presented as percentage of clamp-baseline NEFA results. The insulin sensitivity index (ISI) was calculated as the ratio of SSGIR (μmol/kg/min) to SSIC (μU/mL) according to Mitrakou et al. The insulin elimination from plasma was quantified by calculating the rate constant (k) according to y = a × e−k × t + b, where y represents plasma concentration of insulin (μU/mL) at t, a corresponds to the intersection with the y-axis, b represents clamp-basal plasma concentration of insulin (μU/mL), and t (min) is the sampling time after cessation of the insulin infusion. Hepatic lipid content (TL and TAG) is presented for d0 and d2 in absolute numbers and on d2 as percentage of baseline results of d0.
Statistical analyses were carried out by commercial statistical analysis software.14 Blood test results and data from HEC deviated significantly from normal distribution (Proc UNIVARIATE) and therefore were transformed into the common logarithm before statistical evaluation. Two-factorial analysis of variance for repeated measurements (Proc GLM, REPEATED statement; factor: group, time, and time*group) was used for statistical evaluation of blood test results and milk yield results at d0, d1, d2, and hepatic lipid content at d0 and d2. Within different treatments, means at different time points were compared with baseline results by the Wilcoxon signed-rank test or the paired t-test (hepatic lipid contents and milk yield). Data of the consecutive HEC were tested by 2-factorial analysis of variance for repeated measurements (Proc GLM, REPEATED statement; factor: group, HEC, and group*HEC). BCS was tested for significant differences by the Wilcoxon's 2-sample test.
Blood test results, results of HEC, milk yield, and BCS are presented as medians with range and results of hepatic lipid analysis are presented as means ± SD. To control the comparison-wise error, statistical results of multiple comparisons of means between groups and means within groups to baseline are presented only if the global F-tests were significant (P < .05).
Milk yield and BCS. Neither milk yield nor BCS was significantly different between groups. Milk yield did not change significantly over the study period (Table 2).
|Groups||Group Difference P-Value||Global Effects|
|BCS||0||2.9 (2.50–3.00)||2.8 (2.50–3.25)||Group||.16|
|Milk yield, kg/d||0||23.0 (17–30)||23.0 (15–35)||Group||.80|
|1||21.4 (16–26)||22.5 (14–37)||Time||.39|
|2||22.5 (15–25)||22.0 (14–35)||Timeagroup||.18|
|Glucose, mmol/L||0||3.55 (2.96–4.96)||3.55 (3.13–5.19)||.61||Group||.091|
|1||4.91a (4.44–6.01)||3.51 (2.56–4.63)||.004||Time||.099|
|2||4.51a (3.61–5.42)||4.09 (2.50–4.80)||.27||Timeagroup||.008|
|Insulin, μU/mL||0||3.1 (1.3–9.4)||3.0 (1.6–6.0)||.72||Group||<.001|
|1||7.8a (6.3–11.5)||1.5a (0.9–2.6)||<.001||Time||<.001|
|2||15.7a (12.6–44.5)||6.7 (1.6–30.7)||.046||Timeagroup||.019|
|NEFA, μmol/L||0||950 (562–1920)||525 (344–1754)||.29||Group||.97|
|1||713 (475–1406)||1127a (734–1761)||.076||Time||<.001|
|2||291a (259–1046)||376a (137–522)||.77||Timeagroup||.048|
|BHB, mmol/L||0||0.60 (0.39–1.04)||0.51 (0.32–0.83)||.31||Group||.091|
|1||0.84 (0.39–1.72)||0.62 (0.35–1.18)||.31||Time||<.001|
|2||0.41a (0.34–0.52)||0.23a (0.13–0.32)||.003||Timeagroup||.22|
Blood parameters before and after treatment. At d0 before treatment, mean plasma concentrations of glucose, insulin, NEFA, and BHB were not significantly different between DG-cows and CG-cows (Table 2).
The day after treatment (d1), mean plasma glucose (P = .004) and insulin (P < .001) concentrations were significantly higher in dexamethasone-treated cows compared with CG-cows (Table 2).
Although mean plasma NEFA concentrations decreased steadily from d0 to d2 in DG-cows, NEFA increased in CG-cows on d1 and decreased again at d2 (time*group, P = .048; Table 2).Mean plasma BHB concentrations were not affected by dexamethasone treatment, but decreased in both groups from d0 to d2 (Table 2).
HEC results. Average plasma glucose concentrations, SSGIR, and SSIC are shown in Fig 1. SSGIR was significantly lower (effect: group, P = .007), SSIC significantly higher (effect: group, P = .041; group*HEC, P = .002), and ISI significantly lower (effect: group, P = .012; group*HEC, P = .009) in DG- cows compared with CG-cows (Table 3).
|Groups||Group Difference P-Value||Global Effects|
|SSGIR, μmol/kg/min||I||1.0 (0.9–2.0)||1.4 (0.5–6.6)||.48||Group||.007|
|II||12.2 (7.0–16.8)||21.7 (17.8–27.6)||.002||HEC||<.001|
|III||24.2 (19.1–27.8)||30.7 (20.5–41.5)||.13||Group*HEC||.46|
|IV||25.5 (22.3–32.0)||37.1 (25.9–46.2)||.033|
|V||28.0 (24.7–34.5)||36.0 (25.8–42.0)||.11|
|SSIC, μU/mL||I||8.7 (6.9–10.6)||3.6 (2.9–5.9)||<.001||Group||.041|
|II||31 (18–52)||16 (15–31)||.023||HEC||<.001|
|III||159 (103–224)||116 (107–199)||.48||Group*HEC||.002|
|IV||457 (278–561)||410 (269–459)||.63|
|V||755 (409–987)||650 (410–1161)||.87|
|ISI||I||0.11 (0.09–0.29)||0.35 (0.16–1.73)||.015||Group||.012|
|II||0.47 (0.15–0.78)||1.45 (0.57–1.79)||.004||HEC||<.001|
|III||0.15 (0.09–0.27)||0.24 (0.12–0.36)||.22||Group*HEC||.009|
|IV||0.06 (0.04–0.12)||0.08 (0.06–0.17)||.14|
|V||0.04 (0.03–0.08)||0.05 (0.02–0.08)||.60|
|SSNEFA, μmol/L||I||705 (493–885)||663 (553–924)||.68||Group||.056|
|II||329 (259–518)||256 (180–452)||.14||HEC||<.001|
|III||184 (122–533)||80 (42–412)||.075||Group*HEC||.019|
|IV||162 (125–402)||52 (26–375)||.023|
|V||126 (98–323)||43 (31–392)||.063|
|Rel-SSNEFA,b %||I||106 (50–127)||61 (40–86)||.094||Group||.003|
|II||47 (36–60)||28 (10–35)||.006||HEC||<.001|
|III||30 (17–43)||8 (3–32)||.010||Group*HEC||.041|
|IV||27 (17–34)||4 (3–29)||.003|
|V||19 (14–31)||5 (2–30)||.011|
Mean plasma NEFA concentrations decreased in all cows during insulin infusions (effect: HEC, P < .001). The absolute mean plasma NEFA concentrations were significantly lower after dexamethasone treatment compared with CG only during HEC-IV (P = .023; group*HEC, P = .019), but the decrease in NEFA relative to baseline NEFA concentrations was significantly less pronounced in DG-cows compared with CG-cows (effect: group, P = .003; group*HEC, P = .041; Table 3).
The mean plasma insulin elimination rate constant (k) was 0.0397 ± 0.0077 in DG-cows and 0.0357 ± 0.0066 in CG-cows, and did not differ between groups.
Hepatic TL and TAG content. For hepatic TL (P = .009) and TAG (P = .008) content, significant group by time interactions were found. Relative changes in TL (P = .001) and TAG (P < .001) at d2 compared with d0 were significantly higher in DG-cows than in CG-cows (DG 29 and 42%, CG 4 and 15%, respectively; Table 4).
|Groups||Group Difference P-Value||Global Effects|
|TL mg/g FWa||0||97.2 ± 35.7||95.1 ± 35.7||.92||Group||.55|
|2||66.1 ± 23.3*||91.2 ± 34.9||.17||Time||.002|
|relb||70.6 ± 14.5||96.0 ± 8.4||.001|
|TAG mg/g FW||0||71.0 ± 32.3||72.8 ± 37.0||.92||Group||.53|
|2||42.0 ± 22.3**||63.1 ± 33.3*||.23||Time||<.001|
|relb||57.5 ± 8.9||85.3 ± 10.1||<.001|
Dexamethasone administration induced peripheral insulin resistance as indicated by a decrease in SSGIR required to maintain euglycemia and a decrease in ISI representing the peripheral glucose uptake per insulin unit during HEC. The experimental protocol with 5 consecutive HECs by progressively increasing insulin infusion rates allows differentiation between impaired peripheral insulin sensitivity and response.[17, 18] Because dexamethasone treatment affected ISI significantly only during HEC-I and HEC-II, we propose that the observed peripheral insulin resistance is predominately characterized by decreased insulin sensitivity and less by decreased insulin response (assessed as SSGIR and ISI during the highest insulin infusion rate during HEC-IV and HEC-V).
Compared with CG, SSGIR was decreased in DG-cows 1 day after dexamethasone treatment by approximately 50% during HEC-II and 30% during HEC-IV, which is somewhat less than the 75% decrease found in calves after flumethasone treatment by an insulin infusion rate of 3 mU/kg/min. However, the reported flumethasone-induced decrease in ISI by about 75% in calves was comparable to the results of this study. Differences in SSGIR results may be explained by the different drugs used and the different metabolic situation of calves and lactating cows. Similar results also are reported from horses,[13, 24] humans[10, 25] as well as from in vitro studies using muscle and adipose tissue from in vivo dexamethasone-treated rats.
Studies carried out on rat skeletal muscles and adipocytes demonstrated that the decrease in insulin-stimulated glucose uptake occurs in absence of any consistent effect on insulin receptor number or ligand affinity[27-29] and without decreasing the expression of insulin-dependent glucose transporter (GLUT-4).[26, 30, 31] However, dexamethasone has been shown to impair GLUT-4 translocation to the cell surface.[30-32]
As in other reports,[2, 33] the induced hyperglycemia after dexamethasone-21-isonicotinate treatment is accompanied by an increase in plasma insulin concentration. Hyperinsulinemia is likely a response to hyperglycemia rather than a direct effect of dexamethasone on pancreatic insulin secretion. The increased basal insulin plasma concentrations may explain the significantly higher SSIC in cows after dexamethasone treatment during HEC-I and HEC-II because the insulin elimination rate constant (k) was not significantly different between groups.
The decrease in plasma NEFA concentrations during HEC, which is in line with findings in humans, primarily represents the insulin-mediated decrease in NEFA release from adipose tissue. Hyperinsulinemia decreased plasma NEFA concentrations to approximately 5% of clamp-baseline values in control cows, whereas in dexamethasone-treated cows, NEFA decreased only to approximately 20%. Almost the same result was described in healthy humans after dexamethasone treatment (decrease to 10 and 20% of baseline for untreated and treated subjects, respectively). In vitro studies using human adipocytes assume a direct influence of glucocorticoids on expression of proteins involved in lipolysis, but the results still are inconsistent.[36-38] Glucocorticoids also may affect lipolysis by permissive effects modulating responsiveness to lipolytic hormones, such as catecholamines and growth hormone.[37, 39, 40]
Although dexamethasone may stimulate lipolysis, the missing increase in posttreatment NEFA concentrations in the present and other studies[2, 3, 33] may be explained by concurrent increase in hepatic lipid oxidation and utilization of NEFA by peripheral tissues to meet energy requirements in a state of decreased insulin sensitivity and decreased glucose uptake.
The lack of effect of dexamethasone on mean plasma BHB concentrations in this study is in agreement with results reported previously.[2, 3, 33]
As observed before, dexamethasone treatment induced a significant decrease in both hepatic TL and TAG content, which may be an indication for improved hepatic lipid oxidation or increased TAG release by means of very low-density lipoproteins. In contrast, other studies found no effect of dexamethasone on hepatic TAG content.[2, 33] However, these authors used lower dexamethasone dosages, and cows had lower pretreatment hepatic lipid concentrations.
Inflammatory diseases also may have substantial effects on insulin resistance in cattle. Thus, for this study, cows were only included if free from inflammatory diseases, which might have confounding effects. However, all cows underwent surgical correction of LDA 5 days before the study. Cows with LDA in early lactation already before parturition have increased blood NEFA concentrations compared with healthy herdmates, indicating a predisposition for lipid mobilization.[44, 45] Furthermore, evidence for insulin resistance was reported in cows with LDA.[46, 47] Thus, because it remains unclear if cows with LDA are already initially more disposed to lipid mobilization and insulin resistance, the results of this study should be interpreted with caution in early lactating dairy cows without LDA.
In conclusion, results of our study indicate that treatment with dexamethasone-21-isonicotinate induces insulin resistance in early lactating dairy cows 5 days after correction of LDA, and affects primarily insulin sensitivity rather than insulin response. Dexamethasone impaired the antilipolytic effect of insulin, but did not result in increased plasma NEFA concentrations or hepatic lipidosis. In contrast, dexamethasone appears to decrease hepatic TL and TAG content, possibly by increased hepatic lipid oxidation or release.
Conflict of Interest: Authors disclose no conflict of interest.
Cavafix CertoSplittocan 358, Braun Melsungen AG, Melsungen, Germany
Voren, Boehringer Ingelheim, Germany
Procain-Penicillin-G ad us. vet. aniMedica GmbH, Senden-Bösensell, Germany
bovine insulin 250 mg; Sigma-Aldrich, Germany
Ismatec, REGLO Digital 2 Kanal, Glattbrug ZH, Switzerland
Glucose-Lösung 40% ad us. Vet.; Bela-Pharm GmbH, Vechta, Germany
1 molar KCl, Braun Melsungen
Bard MAGNUM, Biopsy Instrument, Covington, GA Bard MAGNUM, Core Tissue Biopsy Needle, 12 G × 200 mm, BIP GmbH, Türkenfeld, Germany
SSA 370 A, Toshiba, Tokyo, Japan
Procasel 2%, Selectavet, GmbH, Weyarn-Holzolling, Germany
Ascensia, CONTOUR, Bayer Vital GmbH, Leverkusen, Germany
Cobas Mira Plus System from Roche Diagnostic, Mannheim, Germany
Insulin RIA DSL-1600, Texas
SAS, V 9.1, SAS Institute Inc., Cary, NC
- 7Novel minimal invasive technique for measuring hepatic metabolism quantiatvely in dairy cows exemplified by studying hepatic glucose—net production after dexamethasone treatment. In: Chillard Y, Glasser F, Faulconnier Y, eds. Ruminant Physiology: Digestion, Metabolism, and Effects of Nutrition on Reproduction and Welfare. Proceedings of the XIth International Symposium of Ruminant Physiology, Clermont-Ferrand, France. The Netherlands: Wageningen Academic Publishers. 2009;664–666., , , et al.
- 15Intermediary metabolism. In: Van Soest PJ, ed. Nutritional Ecology of the Ruminant. Ithaca, NY: Cornell Univ. Press; 1996:312–324..