Effects of six carbohydrate sources on dog diet digestibility and post-prandial glucose and insulin response

Abstract

The effects of six extruded diets with different starch sources (cassava flour, brewer's rice, corn, sorghum, peas or lentils) on dog total tract apparent digestibility and glycemic and insulinemic response were investigated. The experiment was carried out on thirty-six dogs with six dogs per diet in a completely randomized design. The diets containing brewer's rice and cassava flour presented the greatest digestibility of dry matter, organic matter and gross energy (p < 0.05), followed by corn and sorghum; pea and lentil diets had the lowest. Starch digestibility was greater than 98% in all diets and was greater for brewer's rice and cassava flour than for lentils and peas diets (p < 0.05). Dogs' immediate post-prandial glucose and insulin responses (AUC < or = 30 min) were greater for brewer's rice, corn, and cassava flour diets (p < 0.05), and later meal responses (AUC > or = 30 min) were greater for sorghum, lentil and pea diets (p < 0.05). Variations in diet digestibility and post-prandial response can be explained by differences in chemical composition of each starch source including fibre content and starch granule structure. The nutritional particularities of each starch ingredient can be explored through diet formulations designed to modulate glycemic response. However, more studies are required to support these.

Full text

ORIGINAL ARTICLE
Effects of six carbohydrate sources on dog diet digestibility and
post-prandial glucose and insulin response*
A. C. Carciofi
1
, F. S. Takakura
1
, L. D. de-Oliveira
1
, E. Teshima
1
, J. T. Jeremias
1
, M. A. Brunetto
1
and
F. Prada
2
1 Department of Veterinary Clinic and Surgery, Faculty of Agrarian and Veterinarian Sciences, Sao Paulo State University (UNESP), Jaboticabal,
Brazil, and
2 Faculty of Veterinary Medicine and Animal Science, University of Sao Paulo (USP), Sa
˜o Paulo, Brazil
Introduction
Carbohydrates are the main sources of energy for
many body functions. They are an important compo-
nent of pet foods comprising 30% to 60% of dry
foods and up to 30% of canned formulations. Most
of the carbohydrates found in these products come
from starch. Total tract apparent digestibility of
starch by adult dogs has been reported to be higher
than 95% (Walker et al., 1994; Murray et al., 1999),
although some investigations have demonstrated
that this may change with the type of starch present
in the diet (Belay et al., 1997; Murray et al., 1999;
Twomey et al., 2002). Differences in starch digest-
ibility between sources are because of factors like
cereal type, starch-protein interactions, physical
granule form, starch type, digestion inhibitors, pro-
cessing and particle size after processing (Rooney
and Pflugfelder, 1986; Wolter et al., 1998; Svihus
et al., 2005).
Starch is also the main nutrient responsible for
altering and influencing post-prandial insulin and
glucose responses in both dogs and human beings
(Nguyen et al., 1994; Wolever and Bolognesi, 1996),
and it is known that faster and more complete diges-
tion and absorption of starch leads to greater human
Keywords
canine, digestion, ingredient, meal response,
starch
Correspondence
A. C. Carciofi, Departamento de Clı
´nica e
Cirurgia Veterina
´ria, Faculdade de Cie
ˆncias
Agra
´rias e Veterina
´rias, UNESP, Via de Acesso
Prof. Paulo Donato Castellane, s/n. Jaboticabal
14.884-900 – SP, Brazil. Tel: +55 16 3209
2626; Fax: +55 16 3203 1226; E-mail:
aulus.carciofi@gmail.com
*Presented as part of the 10
th
Congress of
the European Society of Veterinary and
Comparative Nutrition held in Nantes, France,
October 5–7, 2006.
Received: 28 February 2007;
accepted 04 November 2007
First published online: 13 March 2008
Summary
The effects of six extruded diets with different starch sources (cassava
flour, brewer’s rice, corn, sorghum, peas or lentils) on dog total tract
apparent digestibility and glycemic and insulinemic response were inves-
tigated. The experiment was carried out on thirty-six dogs with six dogs
per diet in a completely randomized design. The diets containing
brewer’s rice and cassava flour presented the greatest digestibility of dry
matter, organic matter and gross energy (p < 0.05), followed by corn
and sorghum; pea and lentil diets had the lowest. Starch digestibility
was greater than 98% in all diets and was greater for brewer’s rice and
cassava flour than for lentils and peas diets (p < 0.05). Dogs’ immediate
post-prandial glucose and insulin responses (AUC £30 min) were
greater for brewer’s rice, corn, and cassava flour diets (p < 0.05), and
later meal responses (AUC ‡30 min) were greater for sorghum, lentil
and pea diets (p < 0.05). Variations in diet digestibility and post-prandial
response can be explained by differences in chemical composition of
each starch source including fibre content and starch granule structure.
The nutritional particularities of each starch ingredient can be explored
through diet formulations designed to modulate glycemic response.
However, more studies are required to support these.
DOI: 10.1111/j.1439-0396.2007.00794.x
326 Journal of Animal Physiology and Animal Nutrition 92 (2008) 326–336 ª2008 The Authors. Journal compilation ª2008 Blackwell Publishing Ltd

post-prandial responses (Wolever and Bolognesi,
1996). Other dietetic factors that influence these
responses are diet composition (Nuttall et al., 1984;
Welch et al., 1987; Nishimune et al., 1991; Nguyen
et al., 1998) and processing conditions.
Physiological and pathological states like diabetes
mellitus, obesity, pregnancy, stress, infection, cancer
and advanced age can alter glycemic control (Kahn
et al., 2001). In these situations, the use of a diet
that minimizes – while prolonging – post-prandial
glucose and insulin responses could improve glyce-
mic control and dog welfare (Bouchard and Sunvold,
1999). However, studies on the glucose responses of
potential dog food ingredients are few (Nguyen
et al., 1994; Graham et al., 1994; Bouchard and
Sunvold, 1999).
Therefore the objective of the research reported
here was to investigate the effects of cassava flour,
brewer’s rice, corn, sorghum, peas and lentils on diet
digestibility and the glycemic and insulinemic post-
prandial responses of dogs.
Materials and methods
Animals
Thirty-six neutered mixed-breed dogs, being 21
females and 15 males with body condition scores
between 4 and 6 (Laflamme, 1997), mean body
weight (BW) of 12.5 1.24 kg and mean age of
31.5 years were used in diet digestibility and
meal response tests. The animals were kept in the
Laboratory of Nutrition and Nutritional Diseases at
Sao Paulo State University (Jaboticabal, Brazil). Dur-
ing the digestibility and post-prandial response
experiments the subjects were housed individually
in 1.5 ·1.5 m kennels and had access to fresh water
ad libitum. The Ethics Committee for Animal
Well-Being at the Faculty of Agrarian and Veterinary
Sciences, Sao Paulo State University approved all
experimental procedures.
Diets
In total, six diets were tested and the ingredient
composition of each is reported in Table 1. Each
diet incorporated one of the following carbohy-
drates as its exclusive source of starch: corn,
brewer’s rice, sorghum, peas, lentils, or cassava
flour. Based on chemical compositions of the carbo-
hydrates sources evaluated, additional ingredients
were used to obtain balanced diets containing per-
centages [dry matter (DM) basis] of starch, fat, cal-
cium and phosphorus as similar as possible. Isolated
soybean protein was used to equalize the formula-
tion. All diets contained the same amount of added
salt, vitamins and trace minerals. The amount of
total dietary fibre (TDF) varied according to the
concentration found within the source carbohy-
drate. Chromium oxide (purity 99% w/w; Merck,
Darmstadt, Germany) served as a digestibility mar-
ker and was added to achieve a final concentration
of 3.5 g/kg of diet. Diets were formulated in accor-
dance with the AAFCO (2000) nutrient guide for
dogs, and balanced to meet maintenance require-
ments before being extruded and kibbled under
identical process conditions. The chemical composi-
tion of carbohydrate sources and experimental diets
is shown in Table 2.
Digestibility protocol
The apparent digestibility trial was conducted
according to AAFCO (2000) guidelines for using the
marker method. A 5-day-test-diet adaptation phase
preceded 5 days of faeces collection during the
experimental period. The quantity of diet allotted
was calculated using standard equations that deter-
mine the energy requirements for individual dog
maintenance [maintenance energy (ME in kcal) =
132 ·BW
0.75
kg], in compliance with National
Research Council (1985). Food was weighed each
day, divided into two equal portions and left out at
9am and 5 pm in stainless steel bowls. Bowls were
removed before the next meal and any uneaten food
was weighed and recorded. Faeces were collected
twice daily, weighed, and kept frozen ()15 C) until
analysis. A more detailed description of the proce-
dure for determining apparent digestibility in dogs
with the marker method can be found in Carciofi
et al. (2007).
Faecal samples were scored according to the fol-
lowing system: 1 = watery – liquid that can be
poured; 2 = soft, unformed – stool assumes shape of
container; 3 = soft, formed, moist – softer stool that
retains shape; 4 = hard, formed, dry stool – remains
firm and soft; 5 = hard, dry pellets – small, hard
mass. Faecal pH was determined by mixing 5 ml of
distilled water with 5 g of faeces and measuring the
result with a pH meter (model Q-400-Bd, Quimis,
Brazil).
Post-prandial response tests
Some days after the digestibility trials finished, dogs’
post-prandial glycemic and insulinemic responses
were evaluated following the procedure of Holste
A. C. Carciofi et al. Evaluation of carbohydrate sources for dogs
Journal of Animal Physiology and Animal Nutrition. ª2008 The Authors. Journal compilation ª2008 Blackwell Publishing Ltd 327

et al. (1989), excluding the samples taken at 45 and
90 min and adding one further measurement at
300 min post-meal. Dogs were fed to meet mainte-
nance requirements (ME, kcal = 132 ·BW
0.75
kg;
National Research Council, 1985), and adapted to
their diets for 3 days before testing. During this per-
iod, they were conditioned to ingest all food within
10 min, once per day. Thereby the animals were
tested after 24 h after last meal. Dogs that took
longer than 10 min to consume their total amount
of food were not tested and the procedure was
repeated the following day. On the day of the test,
each dog was aseptically catheterized using a periph-
eral intravenous catheter inserted into the cephalic
vein (Angiocath 20 GA x 1.16 in., Becton-Dickinson,
Franklin Lakes, NJ, USA). Blood samples were taken
pre-feeding (baseline sample, time 0) and 5, 10, 15,
30, 60, 120, 180, 240 and 300 min post-feeding.
Blood was always collected at the same time, begin-
ning at 10 am. Each sample (3 ml) was collected in a
Na-heparin tube, centrifuged (378 gfor 5 min), and
the plasma separated into two Eppendorf tubes.
Plasma samples for glucose measurement were kept
under refrigeration (4 C) for a maximum of 2 h
before analysis; insulin plasma samples were frozen
()70 C) for a maximum of 2 months before they
were analysed.
Laboratory analyses
At the end of the collection period, faeces were
thawed, homogenized, and pooled by dog. Prior to
performing laboratory tests, faeces were dried in a
forced air oven at 55 C for 72 h (320-SE; FANEM,
Sao Paulo, Brazil) and ground in a cutting mill
with a 1-mm sieve. Food samples were ground in
the same way. Diets and faeces were analysed
according to AOAC (1995) standards for DM by
Table 1 Ingredient composition of experi-
mental dog diets
Ingredients
As-fed basis (%)
Cassava
flour Corn Sorghum
Brewer’s
rice Lentil Pea
Cassava flour 42.49 – – – – –
Corn – 53.49 – – – –
Sorghum – – 59.27 – – –
Brewer’s rice – – – 45.66 – –
Lentil – – – – 69.53 –
Pea – – – – – 66.35
Poultry by-product meal 24.00 24.00 23.00 24.00 17.70 19.00
Isolated soybean protein 17.01 9.94 5.16 16.08 0.18 0.09
Poultry fat 5.93 2.00 2.00 3.69 2.02 3.99
Dried whole egg 2.50
Soybean hull 2.00
Brewer’s dried yeast 1.50
Dried hydrolyzed bovine liver 1.50
Dicalcium phosphate 0.90
Calcium carbonate 0.70
Potassium chloride 0.40
Sodium chloride 0.40
Chromium oxide 0.35
Vitamin/mineral premix* 0.10
Mold inhibitor0.10
l-lysine 0.06
dl-methionine 0.05
Antioxidantà0.01
*Per kg of diet: iron, 120 mg; copper, 15 mg; magnesium, 75 mg; zinc, 150 mg; iodine, 2 mg;
selenium, 0.3 mg; vitamin A, 18 000 IU; vitamin D
3
, 1 000 IU; vitamin E, 100 IU; vitamin K, 2 mg;
biotin, 0.6 mg; thiamin, 20 mg; riboflavin, 10mg; pantothenic acid, 50 mg; niacin, 75 mg; vitamin
B
6
, 6 mg; folic acid, 4 mg; vitamin B
12
, 0.1 mg.
Mold Zap

: Ammonium dipropionate, acetic acid, sorbic acid and benzoic acid – Alltech do
Brasil Agroindustrial Limited.
àBanox

: BHA-butilated hydroxyanisole; BHT-butilated hydroxytoluene, propyl gallate and
calcium carbonate – Alltech do Brasil Agroindustrial Limited.
Evaluation of carbohydrate sources for dogs A. C. Carciofi et al.
328 Journal of Animal Physiology and Animal Nutrition. ª2008 The Authors. Journal compilation ª2008 Blackwell Publishing Ltd

oven-drying the sample (934.01), ash through muf-
fle furnace incineration (942.05), crude protein
(CP) applying the Kjeldahl’s method (954.01), acid-
hydrolyzed fat (AHF, 954.02), phosphorus (964.06),
and calcium (968.08). Organic matter (OM) was
calculated by difference (OM = 100)DM). Total die-
tary fibre was measured according to Prosky et al.
(1992) and total starch according to Miller (1959)
and Hendrix (1993). Gross energy (GE) content of
diets and faecal matter was determined using a
bomb calorimeter (Model 1261; Parr Instrument
Company, Moline, IL, USA). Fenton and Fenton
(1979) procedure was used to determine chromium
oxide levels through visible spectrophotometry
(U2010

Hitachi SA, Brisbane, CA, USA). All analy-
ses were carried out in duplicate with a coefficient
of variation below 5%.
Plasma glucose concentrations were determined
by glucose oxidase tests (GOD-ANA; Labtest Diag-
no
´stica S.A., Lagoa Santa, Brazil) using a semi-auto-
mated glucose analyser (Labquest model BIO-2000;
Labtest Diagno
´stica S.A.). Plasma insulin was mea-
sured by radioimmunoassay (RIA) using a commer-
cially available kit (I
125
as tracer, human insulin as
standard; Diagnostic Products Corporation, Los
Angeles, CA, USA) that has already been validated
for dogs (Holste et al., 1989). The intra-assay coeffi-
cient of variation for insulin was 8.4% and the stan-
dard error 0.15 lIU/ml.
Calculations
Apparent digestibility coefficients of DM, OM, CP,
AHF, TDF, GE and total starch were calculated for
each experimental diet using chromium as an
indigestible marker, according to Merchen (1988).
Changes in plasma glucose and insulin concentra-
tions were calculated for each post-prandial period.
Responses were compared for average and maximum
increase, average and maximum incremental increase
(the difference between the absolute glucose or insu-
lin concentration sample and the baseline concentra-
tion), and time to peak increase. The integrated area
under post-prandial glucose and insulin response
curves were calculated using the trapezoidal method.
In addition to measuring total area under the curve
(AUC, from 0 to 300 min), immediate meal response
(AUC £30 min) and later meal response (AUC ‡
30 min) were also calculated. origin (Microcal
Software Version 6.0, OriginLab Corporation, MA,
USA) software was used for AUC computing.
Statistical analyses
The experiment was carried out on thirty-six dogs,
with six dogs per diet in a completely randomized
design. Male and female dogs were distributed
among each group and all groups were evaluated
during the same period. Data were analysed using
Table 2 Chemical composition of starch
sources and experimental dog diets (g/kg dry
matter)* Item
Cassava
flour Corn Sorghum
Brewer’s
rice Lentil Pea
Ingredients
Dry matter (g/kg) 893.7 888.3 887.2 888.8 889.4 893.0
Organic matter 999.3 981.4 978.7 970.7 973 981.1
Crude protein 19.4 105.1 109.3 76 232.6 244.9
Fat 2.5 52.0 60.1 38.0 9.0 11.8
Starch 949.5 784 724.3 881.4 552.6 567.0
Total dietary fibre 15.4 40.7 69.1 16.4 141.3 110.1
Soluble dietary fibre 2.0 6.2 13.8 2.3 29.1 24.7
Insoluble dietary fibre 13.4 34.5 55.3 14.1 112.2 85.4
Crude fibre 33.6 13.2 55.3 8.7 69.3 29.3
Diets
Dry matter (g/kg) 909.5 914.3 913.6 915.7 923.1 912.9
Organic matter 920.2 914.1 922.1 917.1 920.5 918.4
Crude protein 318.4 317.6 282.8 371.0 313.7 295.7
Fat 130.8 118.6 99.9 111.8 91.1 110.0
Starch 417.7 379.9 385.1 399.5 384.8 366.0
Nitrogen-free extract 447.8 445.3 511.2 404.3 489.5 484.0
Total dietary fibre 40.6 94.0 141.3 39.8 159.8 147.4
Crude fibre 23.1 32.6 28.2 30.0 26.2 28.7
*All samples were analysed in duplicate with a coefficient of variation of <5%.
Sorghum containing 0.57% tannin (as-fed base).
A. C. Carciofi et al. Evaluation of carbohydrate sources for dogs
Journal of Animal Physiology and Animal Nutrition. ª2008 The Authors. Journal compilation ª2008 Blackwell Publishing Ltd 329

the general linear model functions of sas (Version
8.0, SAS Institute, Cary, NC, USA). The experimen-
tal unit was dog; the model sums of squares were
separated into treatment (diet) and animal effect. All
multiple comparisons of treatment means were made
using Tukey’s test (p < 0.05). Repeated measures
anova was the statistical method chosen to evaluate
the effects of diet and time on post-prandial plasma
changes. Pairwise mean comparisons were also made
through Tukey’s test (p < 0.05). All data were found
to comply with the assumptions of anova models.
Results
Digestibility and faecal characteristics
Diet chemical compositions are presented in Table 2
and daily nutrient intakes, diet digestibilities and
faecal characteristics are presented in Table 3. All
dogs successfully consumed their experimental diets
and there were no episodes of vomiting, diarrhoea
or meal refusal. Variations in TDF, CP and fat
ingestion between diets were verified (p < 0.05)
and can be explained by the differences in diet
composition. Dogs fed sorghum-based diets ingested
more protein than those receiving lentil-based diets
while ingesting less fat than dogs fed the cassava
flour diet (p < 0.05). Ingestion of other nutrients
was not significantly different between diets
(p > 0.05).
Differences in nutrient digestibility were identified.
Digestibility of DM, OM and GE in brewer’s rice and
cassava flour-based diets was higher than in other
diets (p < 0.05). Protein digestibility was higher in
brewer’s rice than sorghum, corn, pea and lentil
diets (p < 0.05). Fat digestibility was greater for cas-
sava flour and pea diets than other treatments
(p < 0.05). Starch digestibility was >98% for all
treatments with brewer’s rice and cassava flour diets
presenting the greatest digestibility, and pea and len-
til diets the least (p < 0.05). Digestibility of TDF in
pea, lentil and sorghum diets was higher than in
corn and cassava flour treatments (p < 0.05), while
for crude fibre digestibility, no differences were
observed among diets (p > 0.05).
No differences were observed in faecal scores
(p > 0.05). Faecal DM was greater for dogs fed
brewer’s rice, sorghum and corn than for those fed
pea-based diet (p < 0.05). Faecal pH was higher for
the rice-based diet than lentil and pea diets
(p < 0.05).
Table 3 Nutrient intake, apparent total tract digestibility and faecal characteristics of dogs fed experimental diets containing different starch
sources*
Item Cassava flour Corn Sorghum Brewer’s rice Lentil Pea
Daily nutrient intake (g/kg body weight /day)
Dry matter 16.3 0.5 15.2 1.5 14.3 0.5 15.2 1.3 18.9 0.5 16.5 0.2
Organic matter 15.0 0.5 13.9 1.4 13.2 0.4 13.9 1.2 17.4 0.5 15.1 0.2
Crude protein 5.0
bc
0.2 4.8
bc
0.5 4.1
c
0.1 5.6
bc
0.5 5.9
b
0.2 4.9
bc
0.1
Starch 6.8 0.2 5.8 0.6 5.5 0.2 6.1 0.5 7.3 0.2 6.0 0.1
Fat 2.1
b
0.1 1.8
bc
0.2 1.4
c
0.1 1.7
bc
0.1 1.7
bc
0.1 1.8
bc
0.1
Total dietary fibre 1.0
de
0.1 1.4
cd
0.1 1.7
c
0.1 0.9
e
0.1 3.0
b
0.1 2.6
b
0.1
Crude fibre 1.9 0.2 2.5 0.2 2.0 0.1 2.3 0.2 2.5 0.1 2.4 0.1
Apparent digestibility coefficients (%)
Dry matter 83.1
b
0.1 78.6
c
0.5 79.0
c
0.3 82.4
b
0. 74.5
d
0.6 76.1
d
0.7
Organic matter 87.7
b
0.1 83.9
c
0.4 83.8
c
0.2 88.4
b
0.2 79.3
d
0.4 80.1
d
0.7
Crude protein 86.8
bc
0.5 86.1
c
0.3 85.0
cd
0.5 89.0
b
0.5 79.9
e
0.7 83.4
d
0.6
Starch 99.4
b
0.1 99.1
bc
0.1 99.1
bc
0.1 99.3
b
0.1 98.8
c
0.2 98.7
c
0.1
Fat 92.8
b
0.2 89.1
cd
0.3 88.3
d
0.1 89.0
cd
0.4 89.4
cd
0.5 90.0
c
0.2
Gross energy 87.8
b
0.1 84.9
c
0.6 84.2
c
0.4 87.7
b
0.4 78.4
e
0.5 80.9
d
0.8
Total dietary fibre 8.2
d
2.2 11.4
d
3.3 27.0
bc
1.4 17.7
cd
0.7 33.4
b
1.7 30.3
b
3.0
Crude fibre 2.6 0.7 3.1 0.3 3.7 0.3 2.6 0.3 5.0 1.2 4.0 0.8
Faecal characteristics
Faecal score4.2 0.1 3.7 0.1 3.8 0.1 4.2 0.1 3.7 0.1 3.7 0.1
Faecal DM (%) 38.6
cd
1.9 40.2
bc
1.2 40.3
bc
1.6 48.5
b
2.6 33.9
cd
2.7 30.9
d
1.4
pH faecal 6.7
bc
0.1 6.5
bcd
0.1 6.7
bc
0.1 7.1
b
0.1 6.0
d
0.1 6.2
cd
0.2
*Values are means SE of the six dogs per diet.
bcde
Within a row, means without a common superscript differ (p <0.05).
Scores: 1 = watery – liquid that can be poured; 2 = soft, unformed – stool assumes shape of container; 3 = soft, formed moist – softer stool that
retains shape; 4 = hard, formed, dry stool – remains firm and soft; 5 = hard, dry pellets – small, hard mass.
Evaluation of carbohydrate sources for dogs A. C. Carciofi et al.
330 Journal of Animal Physiology and Animal Nutrition. ª2008 The Authors. Journal compilation ª2008 Blackwell Publishing Ltd

Post-prandial glucose and insulin responses
Dogs’ post-prandial glycemic and insulinemic
responses are presented in Fig. 1 and Table 4. Mean
plasma glucose concentrations in dogs given cassava
flour and corn-based diets were significantly lower
than those in dogs consuming the other diets
(p < 0.05). Mean incremental glucose response was
higher in dogs given lentil than pea, cassava flour
and corn diets (p < 0.05). Minimum and maximum
incremental rises in glucose concentration were
lower in dogs consuming pea-based diet than the
other five diets (p < 0.05). Maximum glucose con-
centration occurred within the first hour for cassava
flour, corn, brewer’s rice and peas; these diets stim-
ulated glucose peaks significantly earlier than did
the sorghum diet (p < 0.05). Not all post-prandial
glucose response curves returned to baseline during
the observation period (Fig. 1). At 180 min after
consumption of brewer’s rice, cassava flour and
corn diets, mean plasma glucose concentrations
were not significantly different from basal values
(p > 0.05); conversely, after 300 min the plasma
glucose concentrations of dogs given sorghum, pea
and lentil diets remained above basal values
(p < 0.05).
95
Brewer’s rice Cassava flour
Sorghum
Corn
Pea Lentil
90
85
80
75
70
65
60
55
50
55
50
45
40
35
30
25
20
15
10
5
0
55
50
45
40
35
30
25
20
15
10
5
0
55
50
45
40
35
30
25
20
15
10
5
0
55
50
45
40
35
30
25
20
15
10
5
0
55
50
45
40
35
30
25
20
15
10
5
0
55
50
45
40
35
30
25
20
15
10
5
0
95
90
85
80
75
70
65
60
55
50
95
90
85
80
75
70
65
Plasma glucose
concentration (mg/dL)
Plasma glucose
concentration (mg/dL)
Plasma glucose
concentration (mg/dL)
Plasma insulin
concentration (m UI/mL)
Plasma insulin
concentration (m UI/mL)
Plasma insulin
concentration (m UI/mL)
Plasma glucose
concentration (mg/dL)
Plasma glucose
concentration (mg/dL)
Plasma glucose
concentration (mg/dL)
Plasma insulin
concentration (m UI/mL)
Plasma insulin
concentration (m UI/mL)
Plasma insulin
concentration (m UI/mL)
60
55
50
0 30 60 90 120 150 180 210 240 270 300 0 30 60 90 120150 180210 240270 300
0 30 60 90 120 150 180 210 240 270 300 0 30 60 90 120150 180210 240270 300
0 306090
Minutes after meal feeding
Minutes after meal feeding
Minutes after meal feedin
g
Minutes after meal feedin
g
Minutes after meal feeding
Minutes after meal feeding
120 150
Glucose Insulin
Glucose Insulin
Glucose Insulin Glucose Insulin
Glucose Insulin
Glucose Insulin
180 210 240 270 300 0 30 60 90 120150 180210 240270 300
95
90
85
80
75
70
65
60
55
50
95
90
85
80
75
70
65
60
55
50
95
90
85
80
75
70
65
60
55
50
Fig. 1 Plasma glucose and insulin response curves of dogs fed experimental diets containing different starch sources. Values are means SE of
the six dogs per diet. Values significantly higher than baseline for glucose concentrations (p < 0.05). Values significantly higher than baseline for
insulin concentrations (p < 0.05).
A. C. Carciofi et al. Evaluation of carbohydrate sources for dogs
Journal of Animal Physiology and Animal Nutrition. ª2008 The Authors. Journal compilation ª2008 Blackwell Publishing Ltd 331

Total AUC of glucose was greater in lentil than
cassava flour and corn-based diets (p < 0.05). Imme-
diate post-prandial response (AUC £30 min) was
the greatest for brewer’s rice, intermediate for corn,
lentils, peas and cassava flour, and lowest for sor-
ghum (p < 0.05). Regarding later responses, the
AUC ‡30 min was greater for sorghum, lentil and
pea diets than for brewer’s rice, cassava flour and
corn (p < 0.05).
Mean concentrations of post-prandial insulin were
highest in dogs that had consumed corn-based diet
and lowest for lentil-based diet (p < 0.05). Cassava
flour and brewer’s rice diets induced greater maxi-
mum absolute insulin concentrations than lentils
(p < 0.05), and maximum incremental insulin con-
centration was higher for cassava flour than any
other diet (p < 0.05), although brewer’s rice exhib-
ited intermediate results. Insulin secretion was slow-
est for pea-based diet, which delayed time for
insulin peak in comparison to cassava flour and
brewer’s rice diets (p < 0.05). As Fig. 1 illustrates,
the post-prandial insulin response curve of the sor-
ghum-based diet did not return to baseline during
the 300 min observation period (p < 0.05). Insulin
concentrations had returned to base values by
120 min post-meal for cassava flour diet, 240 min
for brewer’s rice and pea diets, and by 300 min for
corn and lentil diets (p > 0.05).
Total AUC of insulin (0 to 300 min) was smaller
for cassava flour than sorghum (p < 0.05). In the
first 30 min (AUC £30 min) plasma insulin concen-
trations were higher in cassava flour, brewer’s rice
and corn diets than sorghum, lentil and pea diets.
Area under the curve ‡30 min was greater after
ingestion of sorghum-based diet than either brewer’s
rice or cassava flour based diets (p < 0.05).
Discussion
The greater digestibility of brewer’s rice-based diet in
relation to corn verified in the present study was
also described by Belay et al. (1997), Walker et al.
(1994) and Twomey et al. (2002). The use of sor-
ghum as an ingredient in dog formula was evaluated
by Twomey et al. (2002), who found DM digestibil-
ity comparable in diets using sorghum and brewer’s
Table 4. Medium, maximum, and time to peak plasma glucose and insulin concentrations; medium and maximum incremental glucose and insulin
concentrations; and areas under the absolute and incremental glucose and insulin response curves of dogs fed experimental diets containing dif-
ferent starch sources
a
Diet
Cassava flour Corn Sorghum Brewer’s rice Lentil Pea
Glucose (mg/dl)
Mean concentration 77.4
d
0.8 77.8
cd
0.9 80.5
bc
0.7 79.5
bcd
0.9 81.8
b
0.7 79.6
bcd
0.6
Peak 90.1
bc
1.1 91.9
b
1.1 87.2
bc
0.4 91.6
b
0.6 89.2
bc
0.3 86.7
c
2.1
Mean incremental concentration6.0
c
0.8 5.9
c
0.9 8.3
bc
0.7 7.5
bc
0.9 9.73
b
0.7 5.3
c
0.5
Peak incremental concentration 19.9
b
0.6 20.0
b
1.4 15.1
cd
0.9 19.6
b
1.0 16.9
bc
0.97 12.7
d
0.6
Time to peak (min) 27.5
c
2.5 15.0
c
0.1 160.0
b
36.9 15.0
c
0.1 87.5
bc
20.6 55.0
c
5.0
AUC 0–300 minà(mg/dl min) 23027
c
126 22336
c
807 24271
bc
848 23254
bc
246 25254
b
96 24373
bc
378
AUC £30 min§ (mg/dl min) 2461
c
17 2509
bc
26 2304
e
22 2569
b
16 2434
cd
18 2352
d
37
AUC ‡30 min–(mg/dl min) 20566
c
119 20580
c
221 22893
b
115 20685
c
235 22820
b
86 22020
b
341
Insulin (mg/dl)
Mean concentration 18.5
bc
1.7 20.7
b
1.4 20.3
bc
1.25 19.8
bc
1.5 16.7
c
1.2 19.3
bc
1.3
Peak 43.8
b
1.5 36.8
bcd
2.3 35.5
bcd
1.9 41.1
bc
2.6 31.4
d
1.8 35.0
cd
1.8
Mean incremental concentration 14.9 1.8 15.4 1.4 15.5 1.3 15.7 1.6 11.5 1.2 14.2 1.3
Peak incremental concentration 40.2
b
1.2 31.4
cd
1.9 30.7
cd
2.6 37.0
bc
2.5 26.2
d
1.4 29.9
cd
1.6
Time to peak (min) 27.5
c
2.5 37.5
bc
16.8 105.0
bc
21.6 22.5
c
3.3 100.0
bc
28.6 120.0
b
26.0
AUC 0–300 minà(lIU/ml min) 5582
c
418 6428
bc
439 7749
b
245 6284
bc
550 6364
bc
342 7276
bc
311
AUC £30 min§ (lIU/ml min) 772
b
55 781
b
39 491
c
31 768
b
55 442
c
38 504
c
37
AUC ‡30 min–(lIU/ml min) 4810
d
429 5893
bcd
391 7271
b
256 5515
cd
530 5922
bcd
319 6772
bc
298
AUC, area under the curve.
a
Values are means SE of the six dogs per diet.
bcde
Within a row, means without a common superscript differ (p <0.05).
Incremental concentration = absolute glucose or insulin concentration)basal concentration.
àArea under curve 0 to 300 min.
§Area under curve 0 to 30 min.
–Area under curve 30 to 300 min.
Evaluation of carbohydrate sources for dogs A. C. Carciofi et al.
332 Journal of Animal Physiology and Animal Nutrition. ª2008 The Authors. Journal compilation ª2008 Blackwell Publishing Ltd

rice, both being superior to corn-based diet. In a pre-
vious study, however, Twomey et al. (2002) saw les-
ser energy digestibility in sorghum, intermediate
digestibility in corn and greater digestibility in
brewer’s rice. The present research confirmed these
differences between sorghum and brewer’s rice diets,
which can be explained, at least in part, by the addi-
tional quantity of fibre in sorghum. Tannin is
another substance found in sorghum that interacts
with proteins, amino acids, and to a lesser extent
with starch, limiting their digestion (Jansman,
1993). However, the sample used in the present
experiment had such low tannin content (0.57%)
that the presence of the substance did not probably
reduce diet digestibility.
Cassava flour and brewer’s rice within cooked
dog foods were compared by Kamalu (1991), and
as in the present study they encountered similar,
high digestibilities for both diets. No studies on dogs
given pea and/or lentil bean diets were located to
compare and contrast their results, but the lower
digestibilities identified here must be taken into
consideration when creating formulations using
these ingredients. The lower DM digestibilities of
these diets were accompanied by lower protein
digestibilities, verifying the results of other studies
which demonstrated a reduction in protein digest-
ibility in diets with low-digestible carbohydrates
(Murray et al., 1999; Twomey et al., 2002). On
average, pea and lentil diets had TDF digestibilities
2.5 times greater than those of brewer’s rice, corn
and cassava flour. Soluble dietary fibre fractions of
peas and lentils are larger, such that the possibility
exists for greater intestinal fermentation of these
ingredients.
In our investigation, crude fibre was more indi-
gestible than TDF. These results are in assumptions
with others studies about fibre fermentation by dogs,
in vivo (Kienzle et al., 2001) or in vitro models
(Sunvold et al., 1995a,b). Therefore, because of low
digestible nutrients and not veracious amounts of
measurable fibre, crude fibre apparent digestibility
may not be a good method of displaying fibre results
for dogs.
In relation to the starch digestibility of extruded
dog diets, Twomey et al. (2002) also found values
approaching 100%. According to the authors, these
results were because of the extrusion process, which
probably favoured starch gelatinization, making it
almost completely digestible. Bednar et al. (2001)
observed that legumes starches, such as peas and
lentils present reduced digestibility to dogs, a finding
confirmed by the current research. Beyond TDF con-
centration, the authors attributed these differences
between grains and legumes to starch composition;
the proportion of rapidly digestible starch was great-
est in brewer’s rice, corn and sorghum, and lowest
in legumes like peas and lentils.
In the present study, lentil and pea diets produced
lower faecal pH, greater hydrated faeces than rice.
This permits the assumption that for these dogs, a
higher proportion of carbohydrates reached the large
intestine, evidenced by both greater TDF quantities
and lower starch digestibility coefficients, nutrients
which, in turn are made available for bacterial fer-
mentation, according to the assumptions of Cum-
mings and Englyst (1995) and Schu
¨nemann et al.
(1989).
In response to the increase in canine health prob-
lems related to glucose intolerance (e.g. obesity and
diabetes mellitus), an interest has been taken in
investigating diets that favour glycemic control. The
use of starch sources that delay and lengthen glyce-
mic and insulinemic responses while consequently
reducing plasma fluctuations may be beneficial in
these situations (Graham et al., 1994). In this regard,
the present study demonstrated that diets containing
sorghum, lentils or peas could be found to be advan-
tageous over those containing corn, brewer’s rice
and cassava flour.
The interpretation of post-prandial glycemic
responses depends, however, on an integrated evalu-
ation of the diet, one that includes starch-intrinsic
factors like digestion rates and amylose to amylopec-
tin ratios, but also extrinsic influences like ingested
amount, processing and diet composition (Wolever
and Bolognesi, 1996; Heaton et al., 1988; Brand,
1985; Nguyen et al., 1998). In the current study,
diets were formulated to have the closest possible
chemical compositions. Protein intake was sufficient
and variations were small, and according to Nguyen
et al. (1998), differences in the ingestion of this
nutrient do not decisively influence dog post-pran-
dial glucose response, provided that it fulfills the
needs of the animal. Likewise, there were differences
between the fat ingestion of cassava flour and sor-
ghum-based diets. Wolever and Bolognesi (1996)
have already argued that in practical human diets,
the apparent effects of protein and fat on glycemic
response would be negligible. On the other hand,
the quantity of starch ingested by humans corre-
sponds to 46% to 64% of the glycemic variation,
being at times even more important than the type of
starch consumed. In the present study, ingestion of
starch during meal response testing was similar
(p < 0.1) for all dogs.
A. C. Carciofi et al. Evaluation of carbohydrate sources for dogs
Journal of Animal Physiology and Animal Nutrition. ª2008 The Authors. Journal compilation ª2008 Blackwell Publishing Ltd 333

Rice is considered to be a starch source of elevated
glycemic response for humans and canines (Goddard
et al., 1984; Jenkins et al., 1981; Bouchard and
Sunvold, 1999), providing large, rapid alterations in
post-prandial glucose and insulin levels. This charac-
teristic confirmed by the current study is principally
because of the small amount of amylose (Belay
et al., 1997) associated with small quantities of TDF.
The post-prandial response observed in dogs that
consumed cassava flour-based diet also characterized
by rapid rises in glucose and insulin followed by
swift declines back to base levels, may be the result
of an efficient digestibility of extrusion-processed
starch and low-fibre content. Corn diets for dogs
were considered responsible for inducing lower gly-
cemic responses than brewer’s rice (Bouchard and
Sunvold, 1999), a finding that was not confirmed by
this study, as both diets stimulated similar glucose
and insulin responses.
Although corn and sorghum diets presented simi-
lar digestibilities, dogs given these diets experienced
differing post-prandial responses. The major differ-
ences were in relation to time to glucose peak,
which occurred faster for corn, and AUC, which was
greater for corn for the first 30 min and greater for
sorghum for later meal response times. It is worthy
to note that total AUC (0 to 300 min) was similar
for each. According to Rooney and Pflugfelder
(1986), the protein matrix of sorghum’s hard outer
endosperm closely surrounds its starch; this complex
interaction between protein and starch restricts
digestibility.
Dietary fibre content is a further factor that can
alter post-prandial glucose and insulin responses
(Wolever, 1990; Graham et al., 1994). Sorghum,
lentil and pea-based diets provided the highest inges-
tions of TDF, possibly playing a role in delaying and
prolonging the glucose absorption period and in less-
ening the variation in glucose and insulin concentra-
tions demonstrated by these diets in the present
research. Fibre may increase time of gastric emptying
and gastrointestinal transit and diminishes starch
hydrolysis and, consequently, glucose absorption
rate. However, some studies have demonstrated that
in diets with typical fibre levels, variations in the
intake of this ingredient do not significantly influ-
ence post-prandial responses (O’Dea et al., 1980;
Nguyen et al., 1998).
The results reported here indicate that extruded
diets composed of similar ingredients but varying in
starch sources can reveal important differences in
digestibility and post-prandial glycemic and insuline-
mic responses. These differences can be taken into
consideration during the formulation of specialized
products, permitting better technically developed
dog food. Ultimately, long-term studies on the
potential benefits of these carbohydrates are required
for a better understanding of their true effects on
dog health and wellness.
Acknowledgements
The authors acknowledge the financial support of
FAPESP (process 01/08639-3).
References
AAFCO, 2000: Official Publication: Association of
American Feed Control Officials, Atlanta, GA.
AOAC, 1995: Official Methods of Analysis, 15th edn. Associ-
ation of the Official Analytical Chemists, Arlington,
VA.
Bednar, G. E.; Platil, A. R.; Murray, S. M.; Grieshop, C.
M.; Merchen, N. R.; Fahey, G. C. Jr, 2001: Starch and
fiber fractions in selected food and feed ingredients
affect their small intestinal digestibility and ferment-
ability and their large bowel fermentability in vitro in
a canine model. The Journal of Nutrition 131, 276–286.
Belay, T.; Shields-, R. O. Jr; Wiernusz, C. J.; Kigin, P. D.;
Brayman, C. A., 1997: Evaluation of nutrient digestibil-
ity and stool quality of rice (Oryza sativa) based canine
diets. Veterinary Clinical Nutrition 4, 122–129.
Bouchard, G. F.; Sunvold, G. D., 1999: Improving canine
glycemic response to a meal with dietary starch. Pro-
ceedings of the North American Veterinary Conference 1999,
16–19.
Brand, J. C., 1985: Food processing and the glycemic
index. The American Journal of Clinical Nutrition 42,
1192–1196.
Carciofi, A. C.; Vasconcellos, R. S.; de-Oliveira, L. D.;
Brunetto, M. A.; Vale
´rio, A. G.; Bazolli, R. S.; Carrilho,
E. N. V. M.; Prada, F., 2007: Chromic oxide as a digest-
ibility marker for dogs – A comparison of methods of
analysis. Animal Feed Science and Technology 134, 273–
282.
Cummings, J. H.; Englyst, H. N., 1995: Gastrointestinal
effects of food carbohydrate. The American Journal of
Clinical Nutrition 61, 938S–945S.
Fenton, T. W.; Fenton, M., 1979: An improved procedure
for the determination of chromic oxide in feed and
feces. Canadian Journal of Animal Science 59, 631–634.
Goddard, M. S.; Young, G.; Marcus, R., 1984: The effect
of amylose content on insulin and glucose response to
ingested rice. The American Journal of Clinical Nutrition
42, 495–503.
Graham, P. A.; Maskell, I. E.; Nash, A. S., 1994: Canned
high fiber diet and postprandial glycemia in dogs with
Evaluation of carbohydrate sources for dogs A. C. Carciofi et al.
334 Journal of Animal Physiology and Animal Nutrition. ª2008 The Authors. Journal compilation ª2008 Blackwell Publishing Ltd

naturally occurring Diabetes Mellitus. The Journal of
Nutrition 124, 2712–2715.
Heaton, K. W.; Marcus, S. N.; Emmett, P. M.; Bolton, C.
H., 1988: Particle size of wheat, maize, and oat test
meals: effects on plasma glucose and insulin responses
and on the rate of starch digestion in vitro. The Ameri-
can Journal of Clinical Nutrition 47, 675–682.
Hendrix, D. L., 1993: Rapid extraction and analysis of
nonstructural carbohydrates in plant tissues. Crop Sci-
ence 25, 1306–1311.
Holste, L. C.; Nelson, R. W.; Feldman, E. C.; Bottoms, G.
D., 1989: Effect of dry, soft moist, and canned dog
foods on postprandial blood glucose and insulin con-
centrations in healthy dogs. American Journal of Veteri-
nary Research 50, 984–989.
Jansman, A. J. M., 1993: Tannins in feedstuffs for sim-
ple-stomached animals. Nutrition Research Reviews 6,
209–236.
Jenkins, D. J.; Wolever, T. M.; Taylor, R. H.; Barker, H.
M.; Fielden, H.; Baldwin, J. M.; Bowling, A. C.; New-
man, H. C.; Jenkins, A. L.; Goff, D. V., 1981: Glycemic
index of foods: a physiological basis for carbohydrate
exchange. The American Journal of Clinical Nutrition 34,
362–366.
Kahn, S. E.; Prigeon, R. L.; Schwartz, R. S.; Fujimoto, W.
Y.; Knopp, R. H.; Brunzell, J. D.; Porte ., D. Jr, 2001:
Obesity, body fat distribution, insulin sensitivity and
islet b-cell function as explanations for metabolic diver-
sity. The Journal of Nutrition 131, 354S–360S.
Kamalu, B. P., 1991: Digestibility of a nutritionally-bal-
anced cassava (Manihot esculenta Crantz) diet and its
effect on growth in young male dogs. The British Jour-
nal of Nutrition 2, 199–208.
Kienzle, E.; Dobenecker, B.; Eber, S., 2001: Effect of cel-
lulose on the digestibility of high starch versus high fat
diets in dogs. Journal of Animal Physiology and Animal
Nutrition 85, 174–185.
Laflamme, D. P., 1997: Development and validation of a
body condition score system for dogs. Canine Practice
22, 13–18.
Merchen, N. R., 1988: Digestion, absorption and excre-
tion in ruminants. In: D. C. Church (ed.), The Ruminant
Animal: Digestive Physiology and Nutrition. Prentice Hall,
Englewood Cliffs, NJ, pp. 188–189.
Miller, G. L., 1959: Use of dinitrosalicylic acid reagent for
determination of reducing sugar. Analytical Chemistry
31, 426–428.
Murray, S. M.; Fahey, G. C. Jr; Merchen, N. R.; Sunvold,
G. D.; Reinhart, G. A., 1999: Evaluation of selected
high-starch flours as ingredients in canine diets. Jour-
nal of Animal Science 77, 2180–2186.
National Research Council, 1985: Nutrient Requirements of
Dogs. National Academy Press, Washington, USA.
Nguyen, P.; Dumon, H.; Buttin, P.; Martin, L.; Gouro, A.,
1994: Composition of meal influences changes in
postprandial incremental glucose and insulin healthy
dogs. The Journal of Nutrition 124, 2707S–2711S.
Nguyen, P.; Dumon, H.; Biourge, V.; Pouteau, E., 1998:
Glycemic and insulinemic responses after ingestion of
commercial foods in healthy dogs: influence of food
composition. The Journal of Nutrition 128, 2654–2658.
Nishimune, T.; Yakushiji, T.; Sumimoto, T., 1991: Glyce-
mic response and fiber content of some foods. The
American Journal of Clinical Nutrition 54, 414–419.
Nuttall, F. Q.; Moorandian, A. D.; Gannon, M. C.;
Billington, C.; Krezowski, P., 1984: Effect of protein
ingestion on glucose and insulin response to a stan-
dardized oral glucose load. Diabetes Care 7, 465–470.
O’Dea, K.; Nestel, P. J.; Antonoff, L., 1980: Physical fac-
tors influencing postprandial glucose and insulin
responses to starch. The American Journal of Clinical
Nutrition 33, 760–765.
Prosky, L.; Asp, N. G.; Schweizer, T. F.; De Vries, J.
W.; Furda, I., 1992: Determination of insoluble and
soluble dietary fiber in foods and food products: col-
laborative study. Journal of AOAC International 75,
360–367.
Rooney, L. W.; Pflugfelder, R. L., 1986: Factors affecting
starch digestibility with special emphasis on sorghum
and corn. Journal of Animal Science 63, 1607–1623.
Schu
¨nemann, C.; Mu
¨hlum, A.; Junker, S.; Wilfarth, H.;
Meyer, H., 1989: Precaecal and postileal digestibility of
various starches, pH values and concentrations of
organic acids in intestinal chyme. Fortschritte in der
Tierphysiologie und Tiererna
¨hrung 19, 44–58.
Sunvold, G. D.; Fahey, G. C. Jr; Merchen, N. R.; Rein-
hart, G. A., 1995a: In vitro fermentation of selected
fibrous substrates by dog and cat fecal inoculum: influ-
ence of diet composition on substrate organic matter
disappearance and short-chain fatty acid production.
Journal of Animal Science 73, 1110–1122.
Sunvold, G. D.; Fahey, G. C. Jr; Merchen, N. R.; Titge-
meyer, E. C.; Bourquin, L. D.; Bauer, L. L.; Reinhart,
G. A., 1995b: Dietary fiber for dogs IV. In vitro
fermentation of selected fiber sources by dog fecal
inoculum and in vivo digestion and metabolism of
fiber-supplemented diets. Journal of Animal Science 73,
1099–1109.
Svihus, B.; Uhlen, A. K.; Harstad, O. M., 2005: Effect of
starch granule structure, associated components and
processing on nutritive value of cereal starch: a review.
Animal Feed Science and Technology 122, 303–320.
Twomey, L. N.; Pethick, D. W.; Rowe, J. B.; Choct, M.;
Pluske, J. R.; Brown, W.; Laviste, M. C., 2002: The use
of sorghum and corn as alternative to rice in dog foods.
The Journal of Nutrition 132, 1704S–1705S.
Walker, J. A.; Harmon, D. L.; Gross, K. L.; Collings, G. F.,
1994: Evaluation of nutrient utilization in the canine
using the ileal cannulation technique. The Journal of
Nutrition 124, 2672S–2676S.
A. C. Carciofi et al. Evaluation of carbohydrate sources for dogs
Journal of Animal Physiology and Animal Nutrition. ª2008 The Authors. Journal compilation ª2008 Blackwell Publishing Ltd 335

Welch, I. M. C. L.; Hill, B. S. E.; Read, N. W., 1987: Duo-
denal and ileal lipid suppresses postprandial blood glu-
cose and insulin responses in man: possible
implications for the dietary management of diabetes
mellitus. Clinical Science 72, 209–216.
Wolever, T. M. S., 1990: Relationship between dietary
fiber content and composition in foods and the glyce-
mic index. The American Journal of Clinical Nutrition 51,
72–75.
Wolever, T. M. S.; Bolognesi, C., 1996: Prediction of glu-
cose and insulin responses of normal subjects after
consuming mixed meals varying in energy, protein,
fat, carbohydrate and glycemic index. The Journal of
Nutrition 126, 2807–2812.
Wolter, R.; Do Socorro, E. P.; Houdre, C., 1998: Faceal
and ileal digestibility in the dog of diets rich in wheat
or tapioca starch. Recueil de me
´decine ve
´te
´rinaire 174, 45–
55.
Evaluation of carbohydrate sources for dogs A. C. Carciofi et al.
336 Journal of Animal Physiology and Animal Nutrition. ª2008 The Authors. Journal compilation ª2008 Blackwell Publishing Ltd