Dietary supplementation with medium-chain TAG has long-lasting
cognition-enhancing effects in aged dogs
*, Brian Larson
, Joseph A. Araujo
, Winnie Lau
, Christina de Rivera
, Ruben Santana
and Norton W. Milgram
Purina Research, One Checkerboard Square, 2RS, St Louis, MO 63164, USA
CanCog Technologies, Toronto, Canada
Department of Pharmacology, University of Toronto, Toronto, Canada
Department of Psychology, University of Toronto at Scarborough, Toronto, Canada
(Received 20 August 2009 – Revised 23 November 2009 – Accepted 4 January 2010 – First published online 9 February 2010)
The present study focused on the hypothesis that dietary supplementation with medium-chain TAG (MCT) will improve cognitive function in aged
dogs by providing the brain with energy in the form of ketones. Aged Beagle dogs were subjected to a baseline battery of cognitive tests, which
were used to establish cognitively equivalent control or treatment groups. The dogs in the treatment group were maintained on a diet supplemented
with 5·5 % MCT. After an initial wash-in period, all the dogs were tested with a battery of cognitive test protocols, which assessed sequentially
landmark discrimination learning ability, egocentric visuospatial function and attention. The groups were maintained on the diets for 8 months.
The MCT-supplemented group showed signiﬁcantly better performance in most of the test protocols than the control group. The group differences
also varied as a function of task difﬁculty, with the more difﬁcult task showing greater supplementation effects than the easier tasks. The group
given the MCT supplement showed signiﬁcantly elevated levels of b-hydroxybutyrate, a ketone body. These results indicate, ﬁrst, that long-term
supplementation with MCT can have cognition-improving effects, and second, that MCT supplementation increases circulating levels of ketones.
The results support the hypothesis that brain function of aged dogs can be improved by MCT supplementation, which provides the brain with an
alternative energy source.
Brain ageing: Cognitive functions: Dogs: Medium-chain TAG: Ketone bodies
Dog cognitive function, like that of other mammals, becomes
impaired over the course of ageing, and it provides a model of
human cognitive ageing
. Decline in energy metabolism is a
common feature of ageing in animals, and it is one of the sev-
eral processes that are closely associated with age-dependent
cognitive decline. Rapoport et al.
found that brain glucose
metabolism was reduced by up to 30 % between 3 and 12
months of age in rats. London et al.
reported that brain
glucose metabolism was signiﬁcantly reduced in Beagle dogs
at 6 years of age than in 1-year-old dogs. Further changes
occurred later in life, but in a manner that varied between
brain structures. Brain metabolic decline has also been
reported in aged monkeys
and human subjects
, and it
appears to be particularly more pronounced in pathological
ageing. Alexander et al.
reported that cerebral glucose
metabolism was signiﬁcantly lower in old patients with
Alzheimer’s disease than in healthy old control subjects.
Drzezga et al.
traced the development of cognitive decline
in patients with mild cognitive impairment, discovering that
the clinical symptoms of Alzheimer’s disease were associated
with further declines in cerebral glucose metabolism.
These data suggest that the age-associated reduction in
cerebral glucose metabolism is a common feature in ageing,
that the process involved may be progressive, starting
around the middle age, and that metabolic decline contributes
to cognitive decline associated with ageing.
One possible means of counteracting deﬁcits in cerebral
glucose metabolism is by nutritional supplementation.
Although glucose metabolism is the primary source of brain
energy, ketone metabolism provides an alternative pathway,
which normally occurs under starvation conditions. Ketone
bodies are a natural endogenous energy source mainly pro-
duced by the liver from mobilisation of endogenous body fat
and utilised by extrahepatic tissues (brain, heart, kidney,
muscle, etc.). Henderson
has proposed that dietary supple-
mentation with medium-chain TAG (MCT) can be used to
increase levels of ketones in the brain. MCT are converted
to ketone bodies by the liver and, to a lesser extent, by astro-
cytes in the brain. The ketone bodies could then be used by
neurons as an alternative energy source to alleviate the deﬁcit
in glucose metabolism. To partially test this MCT supplemen-
tation hypothesis, Reger et al.
provided an MCT supple-
ment to patients with Alzheimer’s disease and reported an
improvement in cognitive function in a subset of subjects
who were negative for the apo E
4 allele. They also found
that cognitive improvement correlated positively with levels
* Corresponding author: Dr Yuanlong Pan, fax þ 1 314 982 5857, email email@example.com
Abbreviations: BHB, b-hydroxybutyrate; CBC, complete blood count; MCT, medium-chain TAG.
British Journal of Nutrition (2010), 103, 1746–1754 doi:10.1017/S0007114510000097
q The Authors 2010
British Journal of Nutrition
of b-hydroxybutyrate (BHB), a ketone body. More recently,
Page et al.
examined MCT injection in subjects with
type 1 diabetes, and found that MCT were able to reverse
impairment in an array of cognitive domains precipitated by
hypoglycaemia. The present investigation explored the poten-
tial beneﬁcial effects of MCT supplementation in a population
of aged dogs. Following a programme of baseline cognitive
assessment, the dogs were placed into groups that were fed
either a diet containing 5·5 % MCT or a control kibble diet.
Over the course of 8 months on the two diets, the dogs were
tested with a battery of cognitive test protocols that assessed
learning ability, visuospatial function and attention. In the
ﬁrst test protocol, learning ability and visuospatial function
were assessed using a landmark discrimination learning proto-
col. These tasks are intended to assess allocentric spatial abil-
ity, which entails utilisation of external landmarks to localise
objects in space and is distinct from egocentric spatial ability,
in which the subject uses its own body position to identify the
location of external objects. We have previously found that
performance in the landmark protocol is sensitive to age as
well as to a variety of interventions, including maintenance
on an antioxidant diet
and treatment with a combination
of lipoic acid and acetyl-
The second test protocol focused on egocentric spatial abil-
ity, and has been described by Christie et al.
. The protocol
ﬁrst examined the ability of the dog to selectively respond
to an object based on the proximity of the object to its left
or right side, and second, it examined the ability of the dog
to reverse its original response. To assess attention, we used
a modiﬁed protocol for studying oddity discrimination learn-
ing. We have previously described one version of this task,
in which the dogs are trained to select the odd item out of
. In the present investigation, we ﬁrst trained
the dogs to respond to one of two objects, and then examined
the effect of adding additional distractor objects. The task can
be used to assess selective attention because accuracy falls
with an increase in the number of distractors.
Materials and methods
Dogs and housing
The present study used twenty-four Beagle dogs (ten males
and fourteen females) of 7·5–11·6 (9·79 (
SD 0·84)) years of
age at the start of the study. The study protocol was approved
by the CanCog Technologies Institutional Animal Care
Committee, and it followed the guidelines of the Ontario
Ministry of Agriculture. All the dogs had at least 6 months
of previous cognitive test experience, which included
training on a delayed-non-matching-to-position task and an
. Whenever possible, the dogs were group
housed based on compatibility (up to four per pen) in 1·52
by 5·03 m pens. In some cases, the dogs were housed
individually because of compatibility reasons. In such cases,
the pens were divided. The dogs were also provided with
environmental enrichment consisting of toys and beds, and
the opportunity to play outside on a daily basis. Housing
temperature and humidity were held relatively constant by
automated temperature control and continuous ventilation.
Room environmental conditions had design speciﬁcations as
follows: single-pass air supply with 62·3 cubic metres ﬁltered
air changes per minute; relative humidity of 60 ^ 10 %;
temperature of 21 ^ 38C; a natural light – dark cycle.
The control diet was a commercial super premium-type pro-
duct for adult dogs. The test diet was formulated by replacing
5·5 % tallow with 5·5 % MCT. The MCT preparation consisted
of 97 % caprylic acid and 3 % capric acid. Both the diets were
isoenergetic, manufactured by Nestle
Purina PetCare, Inc.
(St Louis, MO, USA), and contained the same levels of pro-
tein, fat and carbohydrates. Dietary ingredient and chemical
composition are provided in Table 1. Diet samples were sent
Purina Analytical Laboratories (Nestle
Petcare) for chemical analyses. Ash, crude fat, crude ﬁbre,
crude protein, moisture and fatty acid proﬁle (linoleic acid,
capric acid and caprylic acid) were measured based on the
Association of Ofﬁcial Agricultural Chemists methods, and
they were found to be 942·05, 922·06, 962·09, 990·03,
930·15 and 996·06, respectively.
The dogs were fed once daily for about an hour. The dogs
were fed to meet their maintenance energy requirements esti-
mated by the formula ‘maintenance energy requirement
¼ 460·6 kJ/d (110 kcal/d) £ body weight
. Dogs had
free access to water via wall-mounted automatic system and/
or water bowls. The dogs were weighed weekly at the begin-
ning of the study, and the food that was provided was adjusted
in order to maintain relatively constant body weights.
Cognitive testing apparatus
The testing apparatus consisted of a wooden box that was
approximately 0·609 m £ 1·15 m £ 1·08 m
in size and was
a modiﬁed version of the Wisconsin General Test Apparatus
Table 1. Ingredients and chemical composition of diets
Ingredients (% as fed)
Plant protein* 31·79 31·79
Animal protein† 20·62 20·62
Cereal grains‡ 30·00 30·00
Tallow 12·5 7·00
MCT 0·00 5·50
Vitamins and minerals 3·24 3·24
Soyabean hulls 1·40 1·40
L-Lys 0·40 0·40
DL-Met 0·05 0·05
Nutrient composition (% as fed)
Moisture 7·41 7·10
Ash 6·32 6·24
Crude protein 32·80 33·10
Crude fat 18·5 18·8
Crude ﬁbre 2·86 2·47
Linoleic acid (% of total fat) 10·1 10·2
Caprylic acid (% of total fat) , 0·10 24·1
Capric acid (% of total fat) , 0·01 1·33
Calculated ME (kJ/g)§ 17·86 18·06
MCT, medium-chain TAG; ME, metabolisable energy.
* Plant protein was obtained from rapeseed, maize and soya.
† Animal protein was obtained from chicken and casein.
‡ Cereal grains included maize, rice and wheat.
§ Calculated based on the predictive equation for ME in dog foods
Medium-chain TAG enhances cognition in dogs 1747
British Journal of Nutrition
widely used in cognitive assessment of primates. The front
contained three height-adjustable gates through which the
dog responded. The experimenter was separated from the
dog by a plastic partition containing a one-way mirror and a
hinged door. The tray was made of Plexiglas and contained
either one medial food well and two lateral food wells, or
four equally spaced food wells, depending on the task.
The food reward was the Pro Plan
Adult Wet Dog Food
Chicken & Rice Entre
Purina Petcare). Approxi-
mately, 1 g of the food constituted each reward, resulting in
a maximum of 12 g of additional food, depending on the task.
Baseline cognitive testing and randomisation
During the baseline phase, all the subjects were tested with a
variable delay version of the delayed-non-matching-to-position
, a size discrimination learning task and a size discrimi-
nation reversal task
. Performance of the subjects in the
three tests was ranked, and the ranking was used to place
the subjects into two cognitively equivalent groups.
Feeding and cognitive testing schedule
One of the groups was then placed on a standard control diet
and the other on a modiﬁed diet containing 5·5 % MCT.
The details about subject distribution and cognitive testing
schedule are outlined in Table 2. After a 1-week wash-in,
all the subjects were tested with a landmark protocol, which
continued for up to 92 d. At 100 d after the initiation of the
study, all the dogs were tested with an egocentric protocol.
The variable object oddity task protocol was started on day
190 and was continued for 35 d (Table 2). For all cognitive
testings, the subjects were tested once daily on one task at a
time. The task order was largely based on previous test proto-
cols. Thus, we have used a similar task order in previous
Landmark discrimination protocol
The landmark discrimination protocol included three separate
tasks: land-0; land-1; land-2
. This protocol started with the
subjects being trained to approach one of two objects based on
their proximity to an external landmark (land-0). The subjects
were tested on successively more difﬁcult versions of the same
general problem. The ﬁrst task, referred to as land-0, utilised a
yellow peg (2 cm £ 2cm£ 9 cm) and two identical white
coasters. The yellow peg was attached to one of the coasters.
In each trial, the experimenter placed the food reward in either
the left or the right food well and positioned the landmark
accordingly. In this and subsequent tasks, food that was inac-
cessible to dogs was placed at the bottom of the coaster associ-
ated with non-reward in order to prevent the dogs from
responding based on olfactory cues. Subjects were able to
obtain food reward if they displaced the coaster attached to
the yellow peg. The dogs were subjected to ten trials per
day, with an inter-trial interval of 30 s. Testing was done
once a day and 6 d per week. The door was raised, and the
tray was moved approximately 25 cm away from the dog for
a brief inspection interval to enable the subject to see the
spatial arrangement on the tray. The tray was then presented
to the dog, and the dog was allowed to respond within 60 s.
Table 2. Cognitive testing schedule, and age range and sex of the dogs
(Mean values with their standard errors)
Age of the dogs (years)
Number of dogs Control MCT
Control MCT All Female Male All Female Male
Days Testing schedule All Female Male All Female Male Mean
SEM Mean SEM Mean SEM Mean SEM Mean SEM Mean SEM
0–6 Wash-in 12 6 6 12 8 4 9·63 0·66 9·67 0·67 9·58 0·70 9·95 1·00 9·85 1·08 10·14 0·91
7– 99 land-0 and land-0 Same as in wash-in Same as in wash-in
land-2 11 6 5 11 7 4 9·65 0·68 9·67 0·67 9·61 0·78 10·00 1·03 9·92 1·15 10·14 0·91
100– 163 Egocentric protocol Same as in wash-in Same as in wash-in
190– 225 Variable object oddity task 11 5 6 9 6 3 9·66 0·68 9·76 0·71 9·58 0·70 9·87 1·05 9·65 1·08 10·32 1·03
Y. Pan et al.1748
British Journal of Nutrition
In this and all subsequent levels, the dogs were required to
respond to the coaster closest to the landmark to obtain food
reward. The correct side was determined randomly by compu-
ter, with the constraint that each side was correct in half of the
trials of each test session. Each dog was allowed a maximum
of thirty test sessions (300 trials) to learn to respond to the
landmark–coaster combination used for land-0 task. A partial
correction procedure was used in which the dogs were per-
mitted to switch their response to the correct object after
responding incorrectly to the coaster alone. Each animal was
allowed only a single correction trial per session.
To move on from the ﬁrst test to the second test (land-1),
the dogs had to complete a two-stage criterion. The ﬁrst
stage was successfully completed when the subject responded
correctly in at least nine of ten trials or in eight of ten trials
over two consecutive days. The second stage was achieved
when the subject responded in at least twenty-one trials
(70 %) over three consecutive sessions. If a dog did not
respond in a trial, it was assigned a score of 0·5, which is
the score that would have been obtained had the dog
responded randomly. In this case, the dog was then given
one extra day of testing to complete the thirty trials, and an
average score of 70 % over all the test days was required to
pass the second stage.
For the land-1 task, the landmark was moved 1 cm medially
and diagonally away from the edge of the coaster. Each dog
was allowed a maximum of thirty test sessions (300 trials)
to learn to respond to a stimulus associated with the landmark
for land-1 task. A partial correction procedure was used in
which the dogs were permitted to correct their response
after making an error once in each session. Dogs that passed
land-1 task were then tested on land-2 task, which was the
same as land-1 task, except that the new landmark position
was diagonally 1 cm away from the edge of the previous
landmark position, and 2 cm from the edge of the coaster.
Dogs underwent ten sessions in land-2 task.
The egocentric protocol had three phases: a preference phase;
an acquisition phase; a reversal learning phase. The egocentric
acquisition protocol evaluates the animals’ spatial learning
ability to use a body-centred coordinate system to locate
objects. The reversal protocol provides an additional measure
of ﬂexibility and executive function
. The preference phase
occurred over a single test day and consisted of presenting the
dog with ten discrete trials with identical objects covering
both the lateral food wells and providing food reward on
each trial. The side chosen most frequently was designated
the preferred side, and it was assigned to be the positive
side for the initial acquisition phase of testing. Thus, if the
dog chose the object to its left most frequently, then the
dog’s left side was designated as its preferred side. For dogs
that did not show a side preference (and responded ﬁve
times to each side), a coin toss was used to determine the
For the acquisition (original learning) phase, the dogs were
tested with twelve trials per session. Each trial consisted of a
single presentation of the stimulus tray with a stimulus cover-
ing a reward on the preferred side lateral well or centre well.
A second non-rewarded stimulus object covered a well
towards the subject’s non-preferred side. Consequently, the
object closest to the dogs’ preferred side was always rewarded.
In any given trial, there were three possible spatial conﬁgura-
tions (left-centre, left-right or right-centre). Each conﬁguration
occurred four times per test session. The original learning
phase was successfully completed when the subject responded
correctly in at least thirty-three of thirty-six trials over three
consecutive test sessions.
All the dogs were subjected to two reversal tests (reversal 1
and reversal 2). The reversal phase was initiated on the day
following the completion of initial learning. The test pro-
cedure was identical to that followed during the acquisition
phase except that the rewarded position was switched to the
opposite side. Thus, if the object closest to a dog’s right
was rewarded in acquisition testing, the object closest to its
left was rewarded in reversal 1 testing. Reversal 1 testing
was successfully completed when the subject responded cor-
rectly in at least twenty-six of thirty-six trials over three
consecutive test sessions. Dogs that passed reversal 1 testing
were then tested with reversal 2 testing, which was identical
to reversal 1 testing, except that the rewarded position was
switched to the opposite side of the reversal 1 testing.
A two-stage learning criterion was used. The ﬁrst stage
required subjects to perform either with at least 90 % accuracy
for 1 d or with at least 80 % accuracy over two consecutive test
days. The second stage required subjects to perform with
above 70 % accuracy over three consecutive days subsequent
to passing the ﬁrst stage criterion. The dogs had to pass the
learning criteria in the ﬁrst phase before they moved on to
the next phase.
Variable object oddity task
This task was developed in order to assess attentional pro-
. The task had three phases: acquisition phase;
‘same distractor’ phase; ‘different distractor’ phase. The ﬁrst
phase examined acquisition of a two-choice discrimination
problem that required the subjects to learn to selectively
respond to a particular object to obtain a food reward.
A two-stage learning criterion was used. The ﬁrst stage
required dogs to perform either with at least 90 % accuracy
for 1 d or with at least 80 % accuracy over two consecutive
test days. The second stage required dogs to perform with
above 70 % accuracy over three consecutive days subsequent
to passing the ﬁrst stage criterion. There were a maximum
of nineteen sessions (one preference test and eighteen acqui-
sition sessions). If a dog did not learn within eighteen sessions,
it was given ﬁve sessions of remedial training and an
additional ten sessions to reach the criterion. If a dog failed
the discrimination, it was not tested further on this task.
Once a dog passed the acquisition phase, it was tested on
the attention task. For the same and different distractor
phases, each trial consisted of presenting to the dogs one,
two, three or four objects, including the object that they had
been trained to respond to during the acquisition phase.
Thus, alternative objects served only as distractors, and the
number of distractors varied from 0 to 3. The same object
was always the one associated with reward in the initial
two-choice discrimination problem. The ﬁrst seven sessions
(1–7) used the same procedure in which the negative stimuli
used in discrimination served as the negative stimuli in the
Medium-chain TAG enhances cognition in dogs 1749
British Journal of Nutrition
attention task. For sessions 8–14, the positive stimulus
remained unchanged and a new distractor was used. In each
test session, the subjects underwent three trials with zero
distractors, three with one distractor, three with two distractors
and three with three distractors. Both accuracy and speed of
responding were used as dependent measures that were indica-
tive of attentional processes.
Body weight, food intake, clinical chemistry,
complete blood count and blood ketone bodies
Before the start of the study, baseline jugular blood samples
were collected for measurements of BHB, CBC and clinical
chemistry. These measurements were repeated after 4 and 8
months of treatment. Blood BHB samples were collected 2 h
after feeding. Samples for clinical chemistry, CBC and BHB
were sent to Advance Vet Lab (Mississauga, Ont., Canada)
for analyses. Food intake was recorded daily, and body
weight was recorded at 2-week intervals.
Errors were used as the dependent measure, and group com-
parisons were made using both Student’s t test and repeated-
measures ANOVA. Values are means with their standard
errors, except the cognitive data in the ﬁgures.
Baseline cognitive tests
Fig. 1 shows that the performance of the two groups, control
and MCT-treated dogs, did not differ in the baseline tests
Effect on performance in the landmark test
The two groups were compared using a repeated measures
ANOVA with total errors in land-0 and land-1 tasks as within-
subject variables and group (supplementation v. control) as
between-subject variable. The analysis revealed a signiﬁcant
effect of group (P¼ 0·015) and task (P¼ 0·0000001), and a
signiﬁcant group-by-task interaction (P¼ 0·043). The origin of
these effects is shown in Fig. 2. The group effect reﬂects that
the MCT group had fewer errors than the control group in
both the tasks. The signiﬁcant group-by-task interaction reﬂects
that the treatment group differed from the control group in the
land-1 task (P¼ 0·02), but not in the land-0 task (P¼ 0·08).
Eleven dogs from each group completed the land-2 task.
The groups were compared using a Student’s t test, which
yielded a signiﬁcant group effect (t(2) ¼ 2·242; P¼ 0·0364).
As illustrated in Fig. 3, the MCT-supplemented dogs per-
formed at a higher level of accuracy than the controls (Fig. 3).
Effect on performance in the egocentric test
The data were analysed with repeated measures ANOVA,
with task (original learning and ﬁrst and second reversals)
as within-subject variable and treatment as between-subject
variable. The results revealed signiﬁcant effects of group
(P¼ 0·01) and task (P, 0·05). As illustrated in Fig. 4, the
task effect reﬂects greater errors during the reversal learning
phase than in initial discrimination learning. In addition,
more accurate learning was found in the second reversal
task than in the ﬁrst. The group effect reﬂects slower learning
by the control group in all the three tasks (Fig. 4). Multiple
comparisons were performed to analyse the group effect (Fish-
er’s least signiﬁcant difference) that revealed signiﬁcant
effects only in the second reversal learning task (P¼ 0·03).
Effect on performance in the variable object oddity test
In the acquisition phase of the attention task, the MCT group
committed fewer errors than the control group, but the differ-
ences were NS (P. 0·05). The attention task was completed
by ten control dogs and nine MCT dogs.
On the single object component, all the dogs performed at a
very high level of accuracy, and the data from this component
were not used in the statistical analysis. The data from 2, 3 and
4 distractor conditions were analysed by repeated measures
ANOVA, with task (same v. different) and number of distrac-
tors (2, 3 or 4) as within-subject variables and treatment as a
Errors to criterion (mean)
Percent accuracy (mean)
Fig. 1. Performance of dogs fed two diets in baseline cognitive tests.
(a) Mean errors to criterion on the size discrimination and reversal learning
tasks. (b) Percentage accuracy on the delayed-non-matching-to-position
task. The data are means with their standard errors, n 12. There were
no statistically signiﬁcant differences. MCT, medium-chain TAG; B, size;
B, size reversal.
Y. Pan et al.1750
British Journal of Nutrition
The results revealed signiﬁcant effects of group (P¼0·018),
task (P¼ 0·003) and number of distractors (P, 0·05). There
were signiﬁcant interactions between task and number of dis-
tractors (P¼ 0·03), and between group and number of distrac-
tors (P¼ 0·0016). These data are shown in Fig. 5, which
illustrate that the dogs made fewer errors in the same task
(Fig. 5(a), when the distractor was the negative object used
in the original training) than in the different task (Fig. 5(b),
when the negative object was originally unfamiliar). Fig. 5
also shows that the group-by-distractors interaction was a
result of the control group performing more poorly than the
treatment group when there were two or three distractors in
both the same and different tasks. Performance was similar
when the groups were only presented with one distractor.
It was also noted that the group effect was larger in the differ-
ent task (Fig. 5(b)) than in the same task (Fig. 5(a)).
Effect on body weight, food intake and blood levels of
-hydroxybutyrate, complete blood count and clinical chemistry
Dogs in the MCT group consumed signiﬁcantly (P, 0·05)
more food during the study, the average daily food intakes
for control and MCT dogs were 144·45 (
SEM 12·94) g and
SEM 12·94) g, respectively. Neither sex nor age signiﬁ-
cantly affected daily food intake within each group. Dogs in
both the groups did not gain signiﬁcant body weight (data
not shown). Table 3 shows that the MCT diet signiﬁcantly
increased blood levels of BHB (which were taken under
non-fasting conditions; P, 0·05).
All the CBC parameters were within normal ranges for
both the groups at baseline and throughout the study (data
not shown). There were signiﬁcant differences between
control and MCT groups in lymphocytes (2·52 (
v. 1·76 (SEM 0·26) £ 10
/litre) and mean cell Hb
concentration (347·00 (
SEM 1·74) v. 340·00 (SEM 1·74) litre/
litre), both of which were higher in the treatment dogs
than in the control dogs at baseline (P, 0·05). Monocytes
SEM 0·05) £ 10
v. 0·12 (SEM 0·05) £ 10
haematocrit (0·45 (
SEM 0·01) v. 0·41 (SEM 0·01) litre/litre)
were signiﬁcantly higher in MCT dogs than in the control
dogs at 120 and 240 d, respectively (P, 0·05). Mean cell Hb
concentration was signiﬁcantly lower in the MCT dogs
SEM 1·74) g/l) than in the control dogs (355·58
SEM 1·74) g/l) at 120 d (P, 0·05). However, all the parameters
of blood clinical chemistry analysis were within the normal
ranges in both the control and the MCT dogs at baseline
and throughout the study (data not shown), with few signiﬁ-
cant differences between the control and MCT dogs. The
MCT dogs had signiﬁcantly (P,0·05) lower blood albumin
SEM 0·75) v. 30·58 (SEM 0·75) g/l), creatinine (47·00
SEM 4·35) v. 62·75 (SEM 4·35) mmol/l) and fasting blood
glucose (3·08 (
SEM 0·17) v. 5·43 (SEM 0·17) mmol/l), and
signiﬁcantly higher blood urea (7·26 (
SEM 0·57) v. 5·60 (SEM
0·57) mmol/l) than the control dogs at the baseline. Fasting
blood glucose (4·39 (
SEM 0·17) v. 4·88 (SEM 0·17) mmol/l)
remained lower in the MCT dogs than in the control dogs at
Errors to criterion
Fig. 2. Effects of dietary medium-chain TAG (MCT) supplementation on
dogs’ performance in object discrimination (land-0) and landmark discrimi-
nation (land-1) learning tasks. The performance was expressed as errors to
criterion. The data are means with their standard errors, n 12. * Mean values
were signiﬁcantly different (P, 0·05) B, MCT; B , control.
Fig. 3. Effects of dietary medium-chain TAG (MCT) supplementation on
dogs’ performance in complex landmark discrimination task (land-2). The data
are means with their standard errors, n 11. The performance was expressed
as total number of errors over ten sessions. * Mean values were signiﬁcantly
different (P, 0·05).
Original learning Rev 1 Rev 2
Fig. 4. Effects of dietary medium-chain TAG (MCT) supplementation on
dogs’ performance in the egocentric discrimination and two subsequent
reversal learning tasks. The performance was expressed as errors to cri-
terion. The data are means with their standard errors, n 12. * Mean values
were signiﬁcantly different (P, 0·05). Rev 1, reversal 1; Rev 2, reversal 2;
B, Controls; B , MCT.
Medium-chain TAG enhances cognition in dogs 1751
British Journal of Nutrition
120 d (P, 0·05). At 240 d, serum alanine aminotransferase
was higher in the MCT dogs (102·83 (
SEM 15·71) U/l) than in
the control dogs (58·17 (
SEM 15·71) U/l; P, 0·05). All thyroid
parameters were within normal ranges in both the MCT-fed
and control diet-fed dogs at baseline and throughout the
study (data not shown).
The primary focus of the present study was to test the hypo-
thesis that dietary supplementation with MCT will have ben-
eﬁcial effects on cognitive function in aged dogs. A group
of aged dogs were ﬁrst subjected to a battery of cognitive
tests to provide baseline data that were used to place the
dogs in two cognitively equivalent groups. One group, the
MCT treatment group, was then placed on a diet containing
a food supplemented with 5·5 % MCT, while the other
group, control group, was fed an isoenergetic diet containing
the same levels of fat, protein and carbohydrate. The results
showed that the MCT-containing diet had positive effects on
cognitive ability and increased serum ketone levels. Collec-
tively, the results support the hypothesis that MCT supplemen-
tation can improve age-related cognitive decline by providing
an alternative source of brain energy that can at least partly
compensate for age-associated decrease in energy metabolism.
The ﬁrst cognitive protocol consisted of three tasks: land-0,
which assessed object discrimination learning, and land-1 and
land-2, which assessed allocentric spatial ability and were
more difﬁcult tasks. Results indicated signiﬁcant differences
between the two groups overall, which were largely driven
by differences in performance in land-1 and land-2 tasks and
provided evidence that MCT supplementation improves
visuospatial function in healthy old dogs.
Marginally signiﬁcant treatment effects were seen in
acquisition of the land-0 task, which was concluded after the
dogs had been on the treatment condition for about 2 weeks,
and signiﬁcant MCT effects were observed within a month
during the feeding trial in acquisition of the land-1 task.
This indicates that short-term administration of MCT sup-
plementation is sufﬁcient to improve learning ability.
The egocentric protocol commenced after approximately 3
months of maintenance on the MCT or control diets. This pro-
tocol had both an initial learning component, which required
an egocentric learning strategy, and a reversal learning,
which provided a measure of executive function and concept
learning. There were signiﬁcant group differences, which
were largely driven by improved reversal learning in the
MCT-treated groups. The greater group differences probably
reﬂect differences in task difﬁculty.
The attention task had three phases. In the ﬁrst phase, the
groups were trained to discriminate between two objects,
and group differences in learning were not found to be signiﬁ-
cant. In the second and third phases, the subjects were pre-
sented with the positive object and from 0 to 3 distractors.
The results indicated that more distractors led to poorer per-
formance, which is consistent with the task providing a
measure of selective attention. Although the groups did not
differ in initial learning, differences occurred in the second
and third phases, suggesting that MCT supplementation
improved the dogs’ ability to focus their attention on the posi-
Collectively, the cognitive assessment data showed that the
performance of the MCT-supplemented group was superior to
that of control group in the land-1 and land-2 tasks, and the
egocentric learning and reversal task and on the variable
object components of the attention protocol. By contrast,
smaller and statistically insigniﬁcant group differences were
noted in the land-0 task, the egocentric discrimination learning
Number of distractors
1·0 2·0 3·0 4·0
1·0 2·0 3·0 4·0
Fig. 5. Effects of dietary medium-chain TAG (MCT) supplementation on
dogs’ performance in two versions of the attention task. (a) Comparison of
the performance of the control and MCT groups when subjects were tested
with the same objects used in the initial training. The Y-axis shows errors,
and the X-axis shows the number of distractors. (b) Comparison of the
performance of the two groups when tested with the same rewarded
object but with different non-rewarded objects. The performance was
expressed as errors to criterion. The data are means with their standard
errors, n 9 for the control group, n 10 for the MCT group. The statistical
analysis revealed a signiﬁcant interaction between number of distractors and
treatments. – X –, Control; ··W··, MCT.
Table 3. Effects of the medium-chain TAG (MCT) diet on blood
b-hydroxybutyrate (BHB) after feeding in old dogs
(Mean values with their standard errors, n 12)
0 120 240
SEM Mean SEM Mean SEM
Serum BHB (mmol/l)
Mean values within a row or column with unlike superscript letters were signiﬁ-
cantly different (P, 0·05).
Y. Pan et al.1752
British Journal of Nutrition
task and the object discrimination phase of the attention task.
A primary difference between the tasks that showed signiﬁ-
cant treatment effects and those that did not is task difﬁculty:
the more difﬁcult tasks were the ones that showed the more
signiﬁcant effects. The observation that the effectiveness of
the intervention varied with task difﬁculty is not surprising,
and may represent a kind of ﬂoor effect. Moreover, we have
previously shown a similar link between task difﬁculty and
a cognition-modifying intervention. Beagle dogs fed an anti-
oxidant-enriched diet showed improved learning of an
oddity discrimination learning task only when the object simi-
larity was increased to make the task more difﬁcult
Serum levels of BHB at baseline and 120 and 240 d after
initiation of the feeding trial were monitored to conﬁrm the
effectiveness of MCT supplementation to increase ketone
levels in ageing dogs. Dogs fed the MCT diet had higher
blood ketones, under fed conditions, conﬁrming the ability of
dietary MCT to increase blood ketone levels without a star-
vation regimen. The resulting ketone levels were well tolerated
by dogs and were well below the ketone levels (0·5 mmol/l)
induced by chronic starvation in dogs
. All the parameters
of CBC and blood biochemistry were within normal ranges at
the end of the study in dogs fed either the MCT or control
diet, indicating that the MCT diet had no adverse effects on
the health of the dogs. The safety of dietary MCT in dogs was
further conﬁrmed by an independent safety study
, in which
diets supplemented with 0, 5, 10 and 15 % MCT were fed to
Beagle dogs for 90 d. The results showed no signs of toxic
effects in dogs at any MCT dose level.
There is a tight coupling between neuronal activity and
cerebral glucose utilisation, and sustaining increased neuronal
activity usually depends on increased ATP production
Normally, most of the ATP production in the neurons
comes from glucose metabolism
. With reduced ability to
metabolise blood glucose, old animals are not able to increase
ATP production high enough to support increased neuronal
activity, which may, at least partially, contribute to the decline
in cognitive function in old animals. In addition, recent evi-
dence also suggests a link between metabolic disorders and
. Since reduction in cerebral glucose
metabolism is also reported in old human subjects
highly likely that dietary MCT supplementation may be able
to improve brain function in old people with and without
any dementia symptoms.
There are also other possible explanations for the positive
effects of MCT supplementation. Loss of PUFA involved
in maintaining neural structure is another consequence of
. Beyond serving as an alternative energy source for
the brain, the cognition-enhancing effects of MCT may be
related to brain distribution and concentration of PUFA
which are involved in maintaining neural structure and known
to decrease during ageing
. This suggestion is supported by
a recent article that revealed increased levels of n-3 fatty acids
in the brain in dogs treated with a programme of MCT
In summary, the present study shows that supplementation
of MCT in ageing dogs can signiﬁcantly increase blood
ketone body concentrations under fed conditions and improve
age-related decline in cognitive function by providing an
alternative source of brain energy in old healthy dogs.
The MCT-containing diet had no adverse effects on CBC
and blood chemistry.
The authors conﬁrm that they have no conﬂicts of interests
associated with the present study. The study was entirely
funded by Nestle
Purina Research. Y. P., B. L., J. A. A.,
C. d. R. and N. W. M. designed the present
study, interpreted the results and prepared the manuscript.
W. L. organised and supported the feeding trial. A. G. and
R. S. contributed to the formulation and production of the
diets, respectively. The authors wish to thank Wendell Kerr
for his statistical analysis of the body weight, food intake
and blood chemical data.
1. Adams B, Chan A, Callahan H, et al. (2000) The canine as a
model of aging and dementia: recent developments. Prog
Neuro-Psychopharmacol Biol Psychiatry 5, 675– 692.
2. Tapp PD & Siwak CT (2006) The canine model of human brain
aging: cognition, behavior and neuropathology. In Handbook of
Models for Human Aging, pp. 415–434. Burlington, MA:
3. Rapoport SI, London ED & Takei H (1982) Brain metabolism
and blood ﬂow during development and aging of the Fischer-
344 rat. Exp Brain Res Suppl 5, 86– 101.
4. London ED, Ohata M, Takei H, et al. (1983) Regional cerebral
metabolic rate for glucose in beagle dogs of different ages.
Neurobiol Aging 4, 121–126.
5. Noda A, Ohba H, Kakiuchi T, et al. (2002) Age-related changes
in cerebral blood ﬂow and glucose metabolism in conscious
rhesus monkeys. Brain Res 936, 76 – 81.
6. Bentourkia M, Bol A, Ivanoiu A, et al. (2000) Comparison of
regional cerebral blood ﬂow and glucose metabolism in the
normal brain: effect of aging. J Neurol Sci 181, 19–28.
7. Alexander GE, Chen K, Pietrini P, et al. (2002) Longitudinal
PET evaluation of cerebral metabolic decline in dementia: a
potential outcome measure in Alzheimer’s disease treatment
studies. Am J Psychiatry 159, 738 – 745.
8. Drzezga A, Lautenschlager N, Siebner H, et al. (2003) Cerebral
metabolic changes accompanying conversion of mild cognitive
impairment into Alzheimer’s disease: a PET follow-up study.
Eur J Nucl Med Mol Imaging 30, 1104– 1113.
9. Henderson ST (2004) High carbohydrate diets and Alzheimer’s
disease. Med Hypotheses 62, 689– 700.
10. Reger MA, Henderson ST, Hale C, et al. (2004) Effects of
b-hydroxybutyrate on cognition in memory-impaired adults.
Neurobiol Aging 25, 311–314.
11. Page KA, Williamson A, Yu N, et al. (2009) Medium-chain
fatty acids improve cognitive function in intensively treated
type 1 diabetic patients and support in vitro synaptic trans-
mission during acute hypoglycemia. Diabetes 58, 1237 –1244.
12. Milgram NW, Head E, Muggenburg B, et al. (2002) Landmark
discrimination learning in the dog: effects of age, an antioxidant
fortiﬁed diet, and cognitive strategy. Neurosci Biobehav Rev 26 ,
13. Milgram NW, Araujo JA, Hagen TM, et al. (2007) Acetyl-
carnitine and alpha-lipoic acid supplementation of aged beagle
dogs improves learning in two landmark discrimination tests.
J FASEB 13, 3756–3762.
14. Christie LA, Studzinski CM, Araujo JA, et al. (2005) A com-
parison of egocentric and allocentric age-dependent spatial
learning in the beagle dog. Prog Neuropsychopharmacol Biol
Psychiatry 29, 361 – 369.
15. Milgram NW, Zicker SC, Head E, et al. (2002) Dietary enrich-
ment counteracts age-associated cognitive dysfunction in
canines. Neurobiol Aging 23, 737– 745.
Medium-chain TAG enhances cognition in dogs 1753
British Journal of Nutrition
16. Chan ADF, Nippak P, Murphey H, et al. (2002) Visuospatial
impairments in aged canines: the role of cognitive-behavioral
ﬂexibility. Behav Neurosci 116, 443– 454.
17. Laﬂamme DP (2007) Five minute veterinary consult: canine and
feline. In Obesity, pp. 982 – 983. Ames, IA: Blackwell Publishing.
18. Milgram NW, Head E, Weiner E, et al. (1994) Cognitive func-
tions and aging in the dog: acquisition of non spatial visual
tasks. Behav Neurosci 108, 57– 68.
19. Head E, Callahan H, Cummings BJ, et al. (1998) Visual-dis-
crimination learning ability and beta-amyloid accumulation in
the dog. Neurobiol Aging 19, 415 –425.
20. Tapp PD, Siwak CT, Estrada J, et al. (2003) Size and reversal
learning in the beagle dog as a measure of executive function
and inhibitory control in aging. Learn Mem 10, 64 – 73.
21. Milgram NW, Adams B, Callahan H, et al. (1999) Landmark
discrimination learning in the dog. Learn Mem 6, 54 – 61.
22. Christie LA, Araujo JA, de Rivera C, et al. (2010) Novel task
for assessing attention in canines: age effects on accuracy and
response latency. Neuroscience (In the press).
23. de Bruijne JJ, Altszuler N, Hampshire J, et al. (1981) Fat mobil-
ization and plasma hormone levels in fasted dogs. Metabolism
24. Matulka RA, Thompson DV & Burdock GA (2009) Lack of
toxicity by medium chain triglycerides (MCT) in canines
during a 90-day feeding study. Food Chem Toxicol 47,
25. Magistretti PJ (2009) Role of glutamate in neuron-glia meta-
bolic coupling. Am J Clin Nutr 90, 875S– 880S.
26. Kodl CT & Seaquist ER (2008) Cognitive dysfunction and
diabetes mellitus. Endocr Rev 29, 494 – 511.
27. Freemantle E, Vandal M, Tremblay-Mercier J, et al. (2006)
Omega-3 fatty acids, energy substrates, and brain function during
aging. Prostaglandins Leukot Essent Fatty Acids 75, 213–220.
28. Taha AY, Ryan MA & Cunnane SC (2005) Despite transient
ketosis, the classic high-fat ketogenic diet induces marked
changes in fatty acid metabolism in rats. Metab Clin Exp 54,
29. Taha AY, Henderson ST & Burnham WM (2009) Dietary
enrichment with medium chain triglycerides (AC-1203) elevates
polyunsaturated fatty acids in the parietal cortex of aged dogs:
implications for treating age-related cognitive decline. Neuro-
chem Res 34, 1619–1625.
30. National Research Council (2006) Nutrient Requirements of
Dogs and Cats. Washington, DC: National Academy Press.
Y. Pan et al.1754
British Journal of Nutrition