Documenti di Didattica
Documenti di Professioni
Documenti di Cultura
DOI 10.1002/ejlt.200700033
Veerle Fieveza
Bruno Vlaemincka
Tom Jenkinsb
Francis Enjalbertc
Michel Doreaud
1 Introduction
Review Article
www.ejlst.com
EFAi0h EFAith
EFAi0h
741
3 Statistical analysis
www.ejlst.com
PUFAi0h PUFAith
PUFAi0h
742
V. Fievez et al.
to derive models that are applicable to all potential studies within a population. Hence, relations and regression
equations presented in the current paper were developed
using a mixed-model regression analysis as described by
St-Pierre [7]. Data were analyzed according to the following model: Yij = B0 1 B1Xij 1 si* 1 bi*Xij 1 eij; with i = the ith
study, j = the jth observation within the ith study, B0 1 B1Xij
= the fixed effect part of the model, si* 1 bi*Xij 1 eij = the
random effect part of the model and with the means of si*
and bi* equaling 0.
To illustrate effects, it was further attempted to graphically
present regression results. However, results from a
mixed-model regression cannot simply be presented as a
Y vs. X graph as the observations come from a (j 1 2)dimensional space, which should first be collapsed into a
two-dimensional space. To account for this loss of
dimensions, adjusted observations should be calculated.
The latter is done by adding the residual from each individual observation to the predicted value of the study
regression [7]. These adjusted observations for study
effects were used to calculate determination coefficients.
Jenkins [18] estimated the average in vivo BH of unprotected PUFA sources from slopes of the regression lines,
estimating ruminal loss of dietary PUFA per unit of PUFA
consumed, after adjustment for the random effect of
study. Slopes were 828 6 17 and 875 6 22 g/kg for
18:2n-6 and 18:3n-3, respectively. However, 18:2n-6 and
18:3n-3 BH data in recent studies are calculated using
data obtained with improved analytical methods allowing
the chromatographic separation of numerous 18:2 and
18:1 isomers. It appears now that 18:2n-6 only represent
a part of the 18:2 fatty acids. Different conjugated
isomers (CLA) and non-conjugated isomers, especially
18:2t11c15, account for a large proportion of the total
18:2, depending on the diet. Loor et al. [19] separated
15 isomers of 18:2 and showed that the percentage of
18:2n-6 in total 18:2 in duodenal chyme varied between
77% for a forage-based diet and 34% for a concentraterich diet supplemented with linseed oil. Obviously, this
might result in a bias in the calculated BH when isomers
are not separated. Indeed, in all determinations of fatty
acid flows published before 2001, one chromatographic
peak was evidenced in the 18:2 area, which was considered as 18:2n-6 whereas it represented in fact a sum of
co-eluted isomers. This explains why in recent studies the
18:2n-6 BH is often higher than in earlier studies (800 and
650 g/kg on average for diets containing less or more
than 60% of concentrate, respectively; [20]). We have
evaluated this bias using 60 dietary treatments from
16 experiments published from 2001 and two unpublished experiments in which the separation between 18:2
isomers was made. The disappearance of 18:2n-6 was
857 g/kg, which is slightly higher than the formerly
www.ejlst.com
743
744
V. Fievez et al.
Tab. 1. Prediction of the 18:2n-6 and 18:3n-3 BH calculated based on total intake and duodenal flows (BH18:2 flow and
BH18:3 flow) from calculations based on (i) 18:2n-6 [BH18:2 (%FA)] and 18:3n-3 [BH18:3 (%FA)] concentrations in total fat
and (ii) proportional to the sum of 18-carbon fatty acids [BH18:2 (%C18) and BH18:3 (%C18)] (n = 218).
Intercept
Estimate
SE
p$
BH18:2flow
NS
BH18:2flow
NS
BH18:3flow
20.101
0.0202
,0.001
BH18:3flow
20.127
0.0239
,0.001
$
#
R2
Independent variable
Variable
Estimate
SE
p$
BH18:2 (%FA)
dietary fat content
BH18:2 (%C18)
dietary fat content
BH18:3 (%FA)
dietary fat content
BH18:3 (%C18)
dietary fat content
0.930
0.009
0.928
0.010
1.075
0.005
1.102
0.006
0.0092
0.0013
0.0105
0.0015
0.0223
0.0009
0.0263
0.0010
,0.001
,0.001
,0.001
,0.001
,0.001
,0.001
,0.001
,0.001
0.896
0.857
0.961
0.970
www.ejlst.com
745
www.ejlst.com
746
V. Fievez et al.
Tab. 2. Overview of characteristics of 24-h batch in vitro incubations used in the current review and indication whether the
study resulted in appropriate (3) simulation or not () of in vivo 18:2n-6 or 18:3n-3 BH and production of the BH end-product
(18:0). The 18:0 cell is empty when the reported results did not allow evaluation of 18:0 production. In the final row, some
characteristics are summarized that impair appropriate in vivo simulation of both 18:2n-6 and 18:3n-3 BH as well as 18:0
production. IND (independent) is mentioned when the whole range of the above-mentioned features ensures appropriate
simulation.
Substrate
Incubated PUFA
main
[2]
18:2
18:3
[3] 18:2
18:3
[4] 18:2
18:3
[26] 18:2
18:3
[27] 18:2
18:3
[33] 18:2
[37]
[40]
[40]
[47]
[51]
Gas
Inoculum
ani- adap- time{{ pre-treatment{
mal## tation{ [h]
pH
start/
24
Total BH 18:0
volume
[mL]
source concentration
[mg/mL]
intro$
type
concentration#
[mg/mL]
seed
0.5 mm
C1U
10
12
0.8-mm sieve
1:1
6.9/UN 120
weight
18.8
1.6-mm sieve
1:1
7.0/6.4 160
hexane evapor. F
16
SH
15
1-mm sieve
1:4
7.0/5.8
25
finely ground
TMR
10
cheese-cloth (2) 1 : 4
UN/UN
52
weight
10
ST
12
cheese-cloth (2) 1 : 4
UN/UN
50
weight
9.6
12
cheese-cloth (2) 1 : 3
UN/UN 208
3
3
3
3
3
3
3
3
3
albumin
tween 80
hexane evapor.
hexane evapor.
weight
grinder
hexane evapor.
none
F
F
F
F
8
8
16
15
3
3
3
3
semi
C
C
C
SH
SH
5
1
1
15
15
particle free
0.5-mm sieve
0.5-mm sieve
1-mm sieve
cheese-cloth (4)
}
1:4
1:4
1:4
1:9
6.8/UN 10
7.1/6.4 50
7.1/6.4 50
7.0/5.7 25
6.7/UN 200
10
SH
15
1-mm sieve
1:4
7.0/6.0
weight
ethanol
UN
tween 80
albumin?
FC
TMR
SM
none
9.6
9.6
10
15
3
3
IND
C
C
ST
UN
3
IND
2
3
12
IND
cheese-cloth (2)
cheese-cloth (2)
cheese-cloth (2)
IND
1 : 4
18:2
1:4
no
buffer
UN/UN 52
UN/UN 52
UN/UN 50
,6.0 IND
oil
oil
corn
oil
oil or
NEFA
NEFA
NEFA
NEFA
oil
oil or
seed
oil
18:2
18:2
18:2
18:2
18:2
18:3
[60] 18:2
18:3
[87] 18:2 oil
[88] 18:2 NEFA
[89] 18:2 oil
Impaired simulation
0.24
0.50
0.31.9
0.04
0.20.6
0.10.5
0.3
0.01
0.03
0.06
0.10.5
0.2
0.040.2
0.040.2
0.32.7
0.2
1.0
0.20.6
0.3
0.30.5
0.10.3
0.7
0.5
dilution{{
50
}}
}}
3}}
3
3
3
3
3
3
3
3
3
3
}}
}}
3}}
3
3
3
In vitro gas accumulation (3) or not (); semi indicates that gas is allowed to accumulate for a certain period (3, 6 or 9 h)
after which it is released.
pH at start (ST) of the incubation and minimum pH after 24 h of incubation; UN indicates values were not reported.
Reducing solution was added together with buffer and inoculum (0.04 of final volume).
}
Washed cell suspensions were added to sterile rumen contents in ratio 3 : 10, no buffer solution added.
}}
6-h [37] and 9-h [40] instead of 24-h in vitro incubation.
www.ejlst.com
747
Fig. 2. Disappearance of 18:2n-6 and 18:3n-3 during 24h batch in vitro incubations with pure unesterified C18:2n6, oils or oilseeds [24, 26, 27, 33, 47, 51, 60, 61, 8789].
Fermentation substrates were mainly hay or lyophilized
grass or alfalfa, but total mixed rations (TMR) have been
added in some experiments n check TMRn. [18:2 lost
(mg/mL) = 0.816(SE = 0.0111; p ,0.001)618:2 initial (mg/mL),
n = 43;
18:3
lost
(mg/mL)
=
R2 = 0.986,
0.882(SE = 0.0054; p ,0.001)618:3 initial (mg/mL), R2 = 0.999,
n = 19].
www.ejlst.com
748
V. Fievez et al.
lipolysis might contribute to inter-experimental differences. Indeed, accumulation of non-esterified EPA [4, 47]
and to a larger extent DHA [26] has been shown to exert a
negative feedback on the BH of EPA and/or DHA.
www.ejlst.com
749
Tab. 3. Overview of extent of in vivo or 24-h in vitro BH of the predominant C18 PUFA of formaldehyde-treated (FT) oil
sources and their experimental control treatment.
Reference
Oil source
Pretreatment
Control
PUFA
(donor)
animal
BH [g/kg]
Con
FT
In vivo
[23]
[45]
[95]
[53]
[50]
linseed
linseed
canola seed
canola/soybean
linseed oil
no
HCOOH
alkaline
emulsification
emulsification
linseed oil
crushed canola seed
yellow grease
linseed oil
18:3
18:3
18:2
18:2
18:3
sheep
sheep
steer
steer
sheep
920
750
714
981$
927
852
648
570
425$
In vitro
[52]
[51]
[51]
[51]
[52]
[52]
[52]
[96]
canola/soybean
linseed
linseed
linseed
canola{
canola/soybean{
canola/soybean
cottonseed oil
no
no
HCOOH
NaOH
unknown
emulsification
emulsification
emulsification
ground linseed
ground linseed
ground linseed
sunflower oil
18:2
18:3
18:3
18:3
18:2
18:2
18:2
18:2
sheep
sheep
sheep
sheep
sheep
sheep
sheep
sheep
850
850
850
960
800
721
482
333
232
200
160
150
Procedure prior to formaldehyde treatment of protein. Emulsification has been mentioned when no information has
been provided on the chemical compounds used to prepare the emulsions.
$
Indicative data, as dietary 18:3 was not reported and has been assumed as 550 g/kg fatty acids, based on average 18:3
concentrations of linseed oil [6] and alfalfa hay [97].
#
Source used in vivo by Sinclair et al. [45].
{
Source used in vivo by Tymchuk et al. [98], without in vivo BH information.
{
Source used in vivo by Zinn et al. [53].
750
V. Fievez et al.
www.ejlst.com
751
Tab. 4. Overview of characteristics of continuous-culture fermentors used in the current review and indication whether the
study resulted in appropriate (3) simulation or not () of in vivo 18:2n-6 or 18:3n-3 BH and production of the BH end-product
(18:0). In the final row, some characteristics are summarized that impair appropriate in vivo simulation of both 18:2n-6 and
18:3n-3 BH as well as 18:0 production. IND (independent) is mentioned when the whole range of the above-mentioned
features ensures appropriate simulation.
Incubated PUFA
main source
Substrate
#
Dilution rate
Inoculum
Adap- pH
tation$
Total BH 18:0
volume
[days]
[mL]
concenintro
tration
[g/kg DMI]
C/F
29
53106
815
57
39
69
28
26
24
2.511
1535
2.3
15
2.3
1530
30
50
emul
emul
50 : 50
24
Soluble carboh. 0.62.4
0 : 100
14
C
ST
C
cheese-cloth (2)
0.068
particle free
0.050.10
cheese-cloth (2) 0.046
0.063
UN
UN
5
6.8
6.5
5.96.3
0 : 10032 : 68 50
0.18
UN
6.87.1 1200
0 : 10032 : 68 50
0.18
UN
6.87.1 1200
0 : 1008 : 92
100
0.10
6.2
1750
emul
62 : 38
120
6.5
1780
emul
62 : 38
120
5.8
1780
emul
feed
IND
50 : 50
24
70 : 30
28
exclusively
IND
soluble carboh.
C
C
UN
cheese-cloth (2)
0.068
cheese-cloth (2)
0.068
particle free
IND
UN
4
IND
6.8
6.8
5.8
700
700
IND
amount
ani- pre-treatment
mal{
[g DM/day]
solid
liquid
[h1]
[h1]
0.12
700
500
700
3
3
3
3
3
3
3
3
3
3
3
3
3
3
3
3
One value for solid and liquid dilution rates indicates single-flow system.
Number of adaptation days of continuous system prior to experimental sampling.
#
Mode of introduction to ensure dispersion of fatty acids in case of oils or NEFA; emul refers to emulsions that are prepared through sonication.
{
Concentrate/forage ratio (C/F); in study [39], soluble carbohydrates (mixture of cellobiose, glucose, maltose and xylose)
were used; in study [65], the concentrate was sucrose.
{
Type of donor animal: steer (ST) or dairy cow (C).
Particle-free inoculum obtained through consecutive filtering through four layers of cheesecloth and centrifugation at
1506g.
$
Overall, a highly significant linear relation between 18:2n6 or 18:3n-3 inputs and their ruminal loss is observed
(Fig. 7). This suggests a negligible effect on the BH of
18:2n-6 and 18:3n-3 of varying continuous-culture conditions, such as the type of the continuous-culture system
(single vs. dual flow), sampling mode (from rumen vessel vs.
effluent), solid (0.040.08 h1) and liquid (0.0630.18 h1)
dilution rates, pH (6.07.25), 18:2n-6 (235 mg/g DMI) or
18:3n-3 (211 mg/g DMI) concentrations, forage/concentrate ratio of the substrate (100 : 030 : 70, wt/wt) and
substrate concentrations (0.10.24 g/100 mL). However,
continuous-culture simulations seem to overestimate in
vivo BH, as suggested from the slope of the linear
regressions (Fig. 7) (948 and 955 g/kg for the 18:2n-6 and
18:3n-3 regression, respectively). Nevertheless, this
might be due to the low number of continuous-culture
studies in combination with the high amounts of unester-
www.ejlst.com
752
V. Fievez et al.
Fig. 7. Disappearance of 18:2n-6 and 18:3n-3 in single(n = 3) and dual-flow (n = 25) continuous-culture fermentors with pure unesterified 18:2n-6, fresh, ensiled or dried
forages [38, 6367]. Substrate concentrations varied between 0.1 and 0.24 g/100 mL. Grey diamond indicates
observation at pH 5.8 [66], which has not been included
in the linear regression. [18:2 lost (g/kg DMI) =
0.948(SE = 0.0115; p ,0.001)618:2 initial (g/kg DMI), R2 = 0.999,
n = 27; 18:3 lost (g/kg DMI) = 0.955(SE = 0.0182; p ,0.001)618:3
initial (g/kg DMI), R2 = 0.995, n = 25].
However, the deviation from the linear curve of this lowpH observation is greater than suggested from the in vivo
meta-analysis (reduction of BH of 109 g/kg for each pH
unit decrease).
tions were not taken into account for the linear regression
development. No 18:1 was detected in gamagrass studied by Eun et al. [63], which resulted in aberrantly low 18:1
intake values when no or low levels of corn were added.
Accordingly, observations of these treatments were
excluded. From the limited number of observations, continuous-culture conditions could be suggested to simulate in vivo conversion of hydrogenated C18 PUFA to 18:0
relatively well. Moreover, this conversion was independent of the continuous-culture conditions, which seems an
improvement compared to the batch in vitro systems.
However, supplementation of fatty acids in continuousculture studies were less than the amounts provoking
accumulation of BH intermediates during in vitro batch
incubations (at least 1.5 g/100 mL substrate in combination with at least 0.5 mg/mL C18 PUFA as NEFA or oils). On
the other hand, fish oil supplementation (2 g EPA 1 DHA/
kg DMI) to continuous cultures with solid dilution rates of
0.03 and 0.06 h1 completely inhibited 18:0 formation
from hydrogenated C18 PUFA of soybean and linseed oil
[62]. Indeed, under these continuous-culture conditions,
the net 18:0 formed (18:0 formed minus oleic acid lost) did
not significantly differ from 0. This would indicate that
continuous cultures are no improvement over batch in
vitro cultures to quantitatively assess the in vivo dose response of fish oil supplementation. However, as the
amount of currently available experimental data from
continuous cultures is limited, further studies are needed
to evaluate the full potential of this methodology.
8 In sacco technique
The in sacco methodology is considered a convenient
compromise between the expensive in vivo approach
www.ejlst.com
753
18:3n-3, 18:2n-6 and production of 18:0 from unprotected fatty acid sources, provided that some considerations are taken into account, which are summarized in the
final rows of Tabs. 2 and 4: (1) The amount of PUFA as
well as the fermentation substrate should be limited to
50 g/kg substrate DM and 10 mg/mL incubation fluid,
respectively. (2) The inoculum should not be treated to
remove all particles, but supplementation of additional
undegradable particulate material is not needed.
(3) Adaptation of the donor animal did not affect in vitro
C18 BH results, but less decisive conclusions could be
made for the type of donor animal, mainly because of the
low number of studies with sheep and steer inoculum.
(4) Weighing, solvent solutions with or without solvent
evaporation, and emulsification through sonication all
seem appropriate techniques to introduce fatty acid
sources into the incubator, and both grinding through a
standard sieve mesh as well as the use of a coffee grinder
are appropriate oil seed pretreatments. (5) Buffer solutions should avoid pH shifts below 6.0, but the inoculum/
buffer ratio is of less importance.
In vitro testing at different pH values is needed to assess
the effectiveness of calcium salts at resisting rumen BH.
However, overestimation of rumen inertness impairs
direct extrapolation of in vitro BH results of technologically protected PUFA sources. Hence, incubation series
of these and other protected sources should include a
negative (unprotected) as well as a positive control of
which in vivo data are available, to enable some speculation on the expected in vivo results.
In vitro BH of EPA and DHA seems dose dependent and
generally underestimates in vivo BH. Moreover, in vitro
supplementation of EPA and DHA most often completely
inhibits 18:0 production, unlike in vivo circumstances
where a dose-dependent inhibition of the trans 18:1 to
18:0 reduction has been suggested. Hence, optimization
of the in vitro methodology to improve simulation of EPA
and DHA rumen metabolism remains challenging. Adaptation of the inoculum donor animal to EPA or DHA supplements should be a first attempt to improve in vitro
simulations.
Acknowledgments
Overall, both batch and continuous-culture in vitro systems show the potential to appropriately simulate BH of
www.ejlst.com
754
V. Fievez et al.
References
[1] G. J. Faichney: Marker methods for measuring digesta flow.
Br J Nutr. 1993, 70, 663664.
[2] F. Akraim, M. C. Nicot, P. Weill, F. Enjalbert: Effects of preconditioning and extrusion of linseed on the ruminal biohydrogenation of fatty acids. 2. In vitro and in situ studies.
Anim Res. 2006, 55, 261271.
[3] A. Troegeler-Meynadier, M. C. Nicot, C. Bayourthe, R. Moncoulon, F. Enjalbert: Effects of pH and concentrations of
linoleic and linolenic acids on extent and intermediates of
ruminal biohydrogenation in vitro. J Dairy Sci. 2003, 86,
40544063.
[4] C. Boeckaert, B. Vlaeminck, J. Mestdagh, V. Fievez: In vitro
examination of DHA-edible micro algae: 1. Effect on rumen
lipolysis and biohydrogenation of linoleic and linolenic acids.
Anim Feed Sci Technol. 2007, 136, 6379.
[5] C. V. D. M. Ribeiro, M. L. Eastridge, J. L. Firkins, N. R. StPierre, D. L. Palmquist: Kinetics of fatty acid biohydrogenation in vitro. J Dairy Sci. 2007, 90, 14051416.
[6] K. J. Harvatine, M. S. Allen: Fat supplements affect fractional
rates of ruminal fatty acid biohydrogenation and passage in
dairy cows. J Nutr. 2006, 136, 677685.
[26] A. A. AbuGhazaleh, T. C. Jenkins: Disappearance of docosahexaenoic and eicosapentaenoic acids from cultures of
mixed ruminal microorganisms. J Dairy Sci. 2004, 87, 645
651.
www.ejlst.com
755
www.ejlst.com
756
V. Fievez et al.
www.ejlst.com