Journal of Insulin Resistance

ISSN: (Online) 2519-7533, (Print) 2412-2785

Page 1 of 9 Original Research

Insulin, glucose and beta-hydroxybutyrate responses

to a medium-chain triglyceride-based sports
supplement: A pilot study

Authors: There is a current trend in endurance sports to move athletes towards a low-carbohydrate
Thomas R. Wood1
diet or use periods of low carbohydrate consumption to increase both health and
Christopher Kelly2
performance. As a result, a market is developing for sports supplements to provide
Affiliations: nutritional support during training and racing for athletes who follow a low-carbohydrate
1
Division of Neonatology, lifestyle. PHAT FIBRE (PF) is a powdered sports supplement that includes medium-chain
Department of Pediatrics,
University of Washington, triglycerides suspended in a digestion-resistant carbohydrate and is tailored to the needs of
United Sates low-carb athletes. Eleven healthy participants were administered 25 g of PF after an
overnight fast. After 30 minutes, median blood glucose increased by 6 mg/dl from 94 mg/dl
Nourish Balance Thrive,
2
to 100 mg/dl ( p = 0.002). At the same time points, median blood beta-hydroxybutyrate
United Sates
(BHB) increased from 0.3 mmol/L to 0.5 mmol/L. The increase in BHB was significant
Corresponding author: ( p = 0.02) after excluding one outlier who had elevated levels of fasting BHB. Insulin levels
Thomas Wood, did not change significantly at any point during the study. In a single participant, a revised
thomasragnarwood@gmail.
com formulation of PF (PFv2) produced a 0.6 mmol/L increase in BHB with no effect on blood
glucose. These data suggest that PF can provide a source of energy for the low-carb athlete
Dates: by supporting ketone production without negatively impacting insulin or blood glucose
Received: 14 Nov. 2016
Accepted: 15 Dec. 2016
levels.
Published: 31 Mar. 2017

How to cite this article:

Wood TR, Kelly C. Insulin,
Introduction
glucose and beta- Over the last decade, there has been a large-scale re-examination, in both the popular and scientific
hydroxybutyrate responses to literatures, about how to best fuel the human body.1,2 As a result, low-carbohydrate diets and
a medium-chain triglyceride-
based sports supplement: A
nutritional ketosis are becoming increasingly considered for the treatment of chronic metabolic
pilot study. J. insul. resist. diseases such as obesity and the metabolic syndrome, as well as neurological diseases such as
2017;2(1), a20. https://doi. Alzheimer’s disease and multiple sclerosis.3,4,5,6 The ‘low-carb’ paradigm has also translated to the
org/10.4102/jir.v2i1.20
fuelling of athletes during endurance exercise. Periods of cyclical low-carbohydrate intake during
Copyright: training can improve time trial performance in triathletes,7 and recent data have shown that ‘keto-
© 2017. The Authors. adapted’ athletes have a dramatically increased ability to metabolise fat aerobically at higher
Licensee: AOSIS. This work percentages of VO2Max intensities.8
is licensed under the
Creative Commons
Attribution License. The fat-adapted athlete is able to rely on body adipose tissue, including intramuscular
triglycerides, to provide fatty acids for aerobic metabolism for longer periods and at higher
exercise intensities.8 However, there is increasing interest in supplements specific to the ‘low-
carb’ (endurance) athlete to support optimal energy metabolism during training or racing. To
be most effective, these supplements should provide calories without significantly affecting
blood glucose or insulin levels in a way that would inhibit endogenous lipolysis and reduce
the capacity for beta-oxidation. One option is to provide a low glycaemic index carbohydrate
source that produces smaller insulin and glucose responses compared to traditional sports
fuels such as maltodextrin, but can still support the maintenance of glycogen levels during
exercise.9 Supplemental fatty acids are also of particular interest to the low-carb athlete,
especially medium-chain triglycerides (MCTs) (6–10 carbon saturated fatty acids). MCTs are
metabolically unique due to their direct absorption into the portal circulation and subsequent
ketogenesis in the liver.10 Ketone bodies increase cardiac efficiency and can be directly used
Read online: by the skeletal muscle as an energy source during aerobic exercise.11,12 Exogenous ketone
Scan this QR supplementation has also been shown to increase endurance cycling performance.13
code with your
smart phone or Supplementing with MCTs results in dose-dependent increases in circulating ketone levels
mobile device (usually measured as beta-hydroxybutyrate, or BHB), but may cause gastrointestinal (GI)
to read online.
distress in larger doses.3

http://www.insulinresistance.org Open Access

Page 2 of 9 Original Research

Anecdotally, the GI effects of MCTs can be reduced when 10 ketone strips to measure blood levels of BHB. A Meridian
they are provided as a powder. However, most commercially Valley Glucose Tolerance Insulin Response (Blood Spot)
available MCT-based powders include high glycaemic index measurement kit (#8069, Meridian Valley Lab, Tukwila,
carbohydrates (i.e. maltodextrin or highly branched cyclic WA) was also included, as was a 25 g pre-weighed pouch of
dextrin),9,14 emulsifiers or caseinate salts. All of these have PHAT FIBRE (PF) (Nourish Balance Thrive, Redding, CA).
potential downsides: (1) high glycaemic index carbohydrates PF is a newly-developed proprietary sports supplement
may increase the insulin response and suppress endogenous powder consisting of MCT oil suspended in digestion-
lipolysis,15 (2) emulsifiers can negatively affect GI health16,17,18 resistant maltodextrin. To minimise the likelihood of
and (3) dairy constituents such as casein are common food adverse reactions to PF, it is packaged in a gluten-free,
allergens.19 dairy-free and peanut-free facility. A 25 g serving of PF
contains 9.4 g carbohydrate (0.2 g sugars, 8.8 g fibre) and
Here, we report on an open-label pilot trial examining the 15.2 g fat. The MCT oil-derived fat consisted of C6:0 (caproic
glucose, insulin and BHB responses to an MCT-based acid, up to 20%), C8:0 (caprylic acid, minimum 50%) and
sports supplement powder using a digestion-resistant C10:0 (capric acid, minimum 30%).
carbohydrate as the sole powder component. The goal is to
develop a powdered MCT source that overcomes the After an overnight fast of at least 8 h (but not more than 16 h),
potential downsides of products currently available on the participants were instructed to follow the manufacturer’s
market. instructions for the Meridian Valley Kraft Assay, using 25 g
of PF dissolved in 250–300 ml of water instead of the 100 g
Methods of dextrose normally used for this test (not included in
the study kit). This assay measures insulin and glucose
Study participants responses over 4 h, determining levels from blood spots on
Healthy participants (n = 11; two women, nine men) were filter paper using enzymatic (glucose) and immuno (insulin)
recruited via email and social media. Full demographic data assays. For this study, blood spots were taken on filter paper
are listed in Table 1. Four of the participants (three men, 1 at baseline, as well as 30 min, 1 h, 2 h, 3 h and 4 h after
women) were elite-level or professional athletes. Median consuming 25 g of PF. Participants were advised to avoid
(range) age was 40 (24–46) years. All of the participants had strenuous exercise and any other caloric intake during the
previous experience with a low-carbohydrate diet or with study period. At each time point, participants also measured
supplements that can boost blood ketone production (MCT and documented their own blood BHB measurements using
oil or exogenous ketones). None of the participants had a the Precision Xtra meter.
history of disease that would interfere with metabolic
responses to a calorie load. All participants performed their Throughout the study, as well as the following day,
own measurements at home according to the study protocol subjective data were collected on satiety, cognitive function,
and gave signed consent for their data to be published in the alertness and GI symptoms, using questions that were
scientific literature. answered on an analogue scale ranging from 1 (not at all) to
5 (very much). A sample of the questions from the study
Experimental protocol protocol is provided in Figure 1. An online pro forma
was provided for each time point to allow data (BHB
Participants received the experimental protocol and a kit
levels, activities over the preceding hour and subjective
containing the study materials through the mail. An
measures) to be collected from the participants in real time.
electronic study protocol was also provided via email. The
An electronic timestamp was automatically recorded
study kit included a Precision Xtra handheld glucose and
when pro formas were submitted to ensure that each
ketone meter (Abbott Diabetes Care Inc., Almeda, CA) with
measurement and the associated data were collected and
recorded in line with the study protocol. Demographic
TABLE 1: Participant demographics.
Participant no. Sex Age Height (cm) Weight (kg) BMI (kg/m2) data, blood BHB levels and subjective data were then
1 M 34 189.0 81.8 22.9 collected directly from the online pro formas. After
2 F 25 171.0 61.0 20.9 allowing the blood spots to dry overnight, participants
3 M 45 170.2 65.9 22.7 shipped their filter papers to Meridian Valley Lab for
4 M 40 175.3 72.7 22.5 glucose and insulin measurements, according to the
5 M 46 185.4 71.8 20.9 manufacturer’s instructions.
6 M 41 180.3 73.5 22.6
7 M 42 160.0 61.4 24.0
8 M 29 182.0 79.0 23.8 Blood glucose measurements
9 M 43 190.5 88.6 24.4
Once blood spot data had been collected, it became apparent
10 M 34 190.5 106.8 29.4
11 F 24 162.6 43.2 16.3
that blood glucose measurements taken from the filter paper
Source: Authors’ own work
were lower than expected, with a median (range) fasting
BMI, Body Mass Index; M, male; F, female. glucose measurement of 70 mg/dl (49 mg/dl–88 mg/dl)

http://www.insulinresistance.org Open Access

Page 3 of 9 Original Research

methods. Within-subject pairwise comparisons of dependent

Over the last hour:
variables (i.e. blood glucose, insulin and BHB) at certain time
1. I have had more gas than usual 1 2 3 4 5
2. I have been in a good mood 1 2 3 4 5 points were compared to the individual’s fasting baseline
3. I have felt more irritable than usual 1 2 3 4 5 measurement using the Wilcoxon signed-rank test. A p-value
4. I have felt full (sasfied) 1 2 3 4 5
5. I have felt shakier than usual 1 2 3 4 5 of < 0.05 was considered statistically significant.
6. I have felt more alert 1 2 3 4 5
7. I have been less able to think clearly 1 2 3 4 5
8. I have been more red (sleepier) than usual
9. I have been thinking more clearly than usual
1
1
2
2
3
3
4
4
5
5
Results
10. I have felt more anxious than usual 1 2 3 4 5 Included data
11. My stomach rumbled more than usual 1 2 3 4 5
12. I have had loose stools 1 2 3 4 5 All participants completed the study in full. Based on the
13. I have been hungry 1 2 3 4 5
14. I have been more easily distracted than usual 1 2 3 4 5
electronic time stamps, mean (standard deviation) time
15. I have been more abdominal pain than usual 1 2 3 4 5 between hourly time points was 60.7 (± 6.2) min, suggesting
16. I have felt focused 1 2 3 4 5
17. My hunger has distracted me 1 2 3 4 5 good adherence to the study protocol. For the duration of
18. I have felt nauseated 1 2 3 4 5 the study, all participants documented normal household
Source: Authors’ own work and work activities that would not be expected to result
Note: Every hour after consuming PF during the study period (4 h), each participant in large changes in insulin, glucose or BHB. One male
answered these 18 questions on a scale of 1 (not at all) to 5 (very much).
FIGURE 1: Subjective measures questionnaire.
participant (#10) had their blood spot card inappropriately
handled by the laboratory, and their glucose and insulin
across the 11 subjects. During the study, two subjects had levels were not available. Metabolic parameters from this
taken simultaneous blood glucose measurements using a participant were therefore excluded from the analysis. Two
handheld glucometer (Precision Xtra; Abbott Diabetes Care other participants (#1 and #9) provided blood spots that
Inc., Almeda, CA), with the glucose level on the handheld were considered low quality by the laboratory due to small
glucometer being on average (median with 95% CI) 26 mg/dl blood volumes. However, the samples were still large
(14 mg/dl–36 mg/dl, n = 12) higher than the blood spot enough to provide data for analysis. In addition, nine
value. One participant also performed the Meridian Kraft participants had at least one insulin level below the
Insulin assay after drinking 100 g of glucose solution, detection limit of the immunoassay (< 2 µU/ml). These
according to the manufacturer’s instructions. Therefore, 18 samples were assigned an insulin level of 1 µU/ml to allow
data points (3 assays, 6 time points per assay) ranging from for statistical analysis.
72 mg/dl to 200 mg/dl (on the handheld meter) were
available to compare values taken from the blood spot filter Subjective measures
paper assay with a blood glucose measurement taken at the
None of the participants reported a significant effect of PF on
same time as the blood spot was collected. Linear regression
irritability or negative affect: shakiness, sleepiness, distractibility,
analysis showed that the two were significantly correlated
anxiety, loose stools, abdominal pain or nausea. Median
(R2 = 0.83, p < 0.0001), so the resulting regression equation
hunger score was 1 (‘not at all’ hungry) for the first 2 h of the
was used to create a ‘calibration curve’ and calculate
study, increasing to a median score of 2 (‘a little bit’ hungry)
estimated blood glucose measurements from blood spot
at 3 h and 4 h (Figure 2a). This corresponded with a median
results for all participants.
satiety (fullness) score of 3 (‘somewhat’ full) at 1 h and 2 h,
which decreased to a score of 2 (‘a little bit’ full) at 3 h and 4 h
Product development (Figure 2b). Participants reported a median focus score of 3
As part of ongoing pilot experiments, one participant (#4) (‘somewhat’ focused) throughout the entire study period
repeated the study using an updated formulation of PHAT (Figure 2d).
FIBRE (PFv2), measuring both blood glucose and BHB using
a handheld glucose and ketone meter, as described above. Glucose responses to PHAT FIBRE
PFv2 was formulated to include 70% C8:0 (caprylic acid)
Based on the data reported by the laboratory, median (95%
triglycerides by weight, suspended in 30% digestion-resistant
CI) glucose at baseline was 70 mg/dl (49 mg/dl–80 mg/dl).
maltodextrin.
Blood glucose increased by 9 mg/dl 30 min after consuming
PF to 79 mg/dl (61 mg/dl–103 mg/dl), which was statistically
Statistical analysis significant (p = 0.002). However, as described in the methods
Statistical analyses were performed using SPSS software above, it appeared that the enzymatic assay tended to
version 22 (SPSS Inc., Chicago, IL) and GraphPad Prism underestimate blood glucose values, so a linear regression
version 7 (GraphPad Software, La Jolla, CA). For subjective equation was used to adjust the results based on a subset of
measures, answers were converted to their numerical data for which a blood glucose measurement was also
equivalent for graphical representation and statistical available from the handheld glucometer (Figure 3). Blood
analysis. Metabolic data did not consistently follow a normal glucose results were therefore adjusted based on the equation:
distribution and were therefore analysed using nonparametric Y = 0.6927*X + 45.04.

http://www.insulinresistance.org Open Access

Page 4 of 9 Original Research

5 a 5 b 20 * a

Glucose change (baseline to 60 minuts,

Hunger score

Saety score
4 4
3 3 15

2 2 10
1 1
5

mg/dl)
1 2 3 4 1 2 3 4
Time (hours) Time (hours) 0
0 30 60
5 c 5 d -5 Time (minutes)
Alertness score

4 4
Focus score

-10
3 3
2 2 -15

1 1 140 b
*
1 2 3 4 1 2 3 4
Time (hours) Time (hours)

Blood Glucose (mg/Id)

120
Source: Authors’ own work
Note: None of the participants reported a significant effect of PF on irritability, mood,
shakiness, sleepiness, distractibility, anxiety, loose stools, abdominal pain or nausea. Plots
display median (range): a-d 100
FIGURE 2: Subjective measures of mood, satiety and hunger: (a) score for
hunger, (b) satiety, (c) feeling more alert and (d) focus.
80

250
Handheld glucometer glucose (mg/dl)

60
200 0 0.5 1 2 3 4
Time (hours)

150 Source: Authors’ own work

FIGURE 4: Blood glucose responses to PF. (a) Adjusted individual blood glucose
changes (from baseline) at 30 min and 60 min. The magnitude of blood glucose
100 change at 30 min ranged from -2.8 mg/dl to +16.0 mg/dl. In 70% of participants,
glucose had returned to baseline by 60 min. (b) Scatter plot (line at median) of
blood glucose at baseline (fasting) and across the 4-h study period. Median
50 Y = 0.6927*× + 45.04 (95% CI) blood glucose after 30 min had increased from 94 mg/dl (79 mg/dl–101
mg/dl) to 100 mg/dl (87 mg/dl–116 mg/dl), which was statistically significant
( p = 0.002). Blood glucose returned to baseline by 1 h and remained similar to
0 baseline for the remainder of the experimental period.
50 100 150 200
Blood spot assay glucose (mg/dl) insulin fell to 1 µU/ml (1.0 µU/ml–6.1 µU/ml) by 2 h and
Source: Authors’ own work
remained at 1 µU/ml for the remainder of the study period.
FIGURE 3: Blood glucose calibration. Linear regression of 18 data points where Individual insulin responses for the first hour, and aggregated
data were available for both the blood spot enzymatic assay and a handheld data over the 4-h study period, are shown in Figure 5.
glucometer. The single point at 200 mg/dl is from one participant (#4) who had
data using the same assay after a 100 g glucose load. A significant correlation
between the two (R2 = 0.83, p < 0.0001) was seen, and the resulting equation
(Y = 0.6927*X + 45.04) was used to adjust glucose measurements for all Beta-hydroxybutyrate responses to PHAT FIBRE
participants in the analysis.
Median (95% CI) BHB at baseline was 0.3 mM (0.2 mM–0.5
mM), which increased by 0.2 mM to 0.5 mM (0.2 mM–0.8
For the adjusted blood glucose results, median (95% CI)
mM) after 30 min. However, this difference was not
glucose at baseline was 94 mg/dl (79 mg/dl–101 mg/dl).
statistically significant ( p = 0.07). One outlier (participant #8)
This increased by 6 mg/dl to 100 mg/dl (87 mg/dl–116 mg/
had a fasting BHB of 1.1 mM, which fell to 0.9 mM after 30
dl), which was statistically significant ( p = 0.002). Blood
min. Excluding this outlier (leaving 10 participants who had
glucose returned to baseline by 1 h and remained similar to
a fasting BHB of 0.2 mM–0.5 mM), PF significantly increased
baseline for the remainder of the experimental period.
BHB at 30 min compared to baseline ( p = 0.02). Including all
Individual glucose responses for the first hour, and
aggregated data over the 4-h study period, are shown in participants, median BHB decreased to 0.4 mM (0.2 mM–0.6
Figure 4. mM) at 1 h, but increased again to 0.5 mM (0.4 mM–0.6 mM)
at both 2 h and 3 h and 0.45 mM (0.4 mM–0.6 mM) at 4 h. At
both 3 h and 4 h, BHB was significantly elevated compared to
Insulin responses to PHAT FIBRE fasting baseline levels ( p = 0.03 and p = 0.002, respectively).
Median (95% CI) insulin level was 2.1 µU/ml (1.0 µU/ml– Individual BHB responses for the first hour, and aggregated
7.6 µU/ml) at baseline, which increased by 4.6 µU/ml to data over the 4-h study period, are shown in Figure 6. No
6.7 µU/ml (1.0 µU/ml–8.9 µU/ml) after 30 min. However, correlation between degree of glucose or insulin response
this difference was not statistically significant. Median and degree of BHB response was seen.

http://www.insulinresistance.org Open Access

Page 5 of 9 Original Research

15 a 1.2 a

1.0
#

Blood ketones (mmol/L)

10
Insulin (µU/ml)

0.8

0.6

5
0.4

0.2

0 0.0
0 30 60 0 30 60
Time (hours) Time (minutes)
15 b * b

1.5
#
10
Insulin (µU/ml)

Blood ketones (mmol/L) 1.0

0.5

0
0 0.5 1 2 3 4
Time (hours) 0.0
0 0.5 1 2 3 4
Source: Authors’ own work
Time (hours)
FIGURE 5: Insulin responses to PF. (a) Individual insulin changes (from baseline)
at 30 min and 60 min. The magnitude of insulin change at 30 min ranged from Source: Authors’ own work
-2.5 µU/ml to +11.0 µU/ml. (b) Scatter plot (line at median) of insulin levels at
baseline (fasting) and across the 4-h study period. Median (95% CI) insulin after FIGURE 6: Beta-hydroxybutyrate responses to PF. (a) Individual BHB changes
30 min had increased from 2.1 µU/ml (1.0 µU/ml–7.6 µU/ml) to 6.7 µU/ml (from baseline) at 30 min and 60 min. The magnitude of BHB change at 30 min
(1.0 µU/ml–8.9 µU/ml). However, this change was not significantly different ranged from -0.2 mmol/L to +0.6 mmol/L. (b) Scatter plot (line at median) of
(Wilcoxon signed rank test, p = 0.2). insulin levels at baseline (fasting) and across the 4-h study period. Median (95%
CI) BHB after 30 min had increased from 0.3 mmol/L (0.2 mmol/L–0.5 mmol/L)
to 0.5 mmol/L (0.2 mmol/L–0.8 mmol/L). At 4 h, median BHB was 0.45 mM (0.4
mM–0.6 mM). #Indicates significantly increased BHB ( p = 0.02) at 30 min
Glucose and beta-hydroxybutyrate responses compared to baseline, after exclusion of a single outlier (indicated by open grey
to PFv2 circles in a). *Indicates whole-group significant increase in BHB at 3h and 4 h
compared to baseline ( p = 0.3 and p = 0.002, respectively).
In a single participant (#4), 25 g of PFv2 did not elevate blood
glucose from a baseline of 88 mg/dl. This is in contrast to a well other high-profile medical doctors such as Timothy
large 25 mg/dl glucose response seen in this participant after
Noakes and Philip Maffetone.1,22,23 As such, there is increasing
25 g of PFv1, from a baseline of 89 mg/dl to 114 mg/dl at 30
interest in supplements that can be used to fuel prolonged
min (Figure 7a). A greater BHB response was seen with PFv2
endurance efforts without large increases in blood glucose
compared to PFv1, increasing from 0.2 mM at baseline to 0.7
and insulin that would suppress endogenous fatty acid
mM after 30 min, reaching a peak of 0.8 mM at 2 h and
oxidation. Here, we show that 25 g of a powdered MCT oil-
remaining above 0.5 mM for the whole study period
based sport supplement (PF) in resting healthy participants
(Figure 7b).
can significantly increase blood BHB levels with a less than
10% increase in blood glucose and without significantly
Discussion affecting insulin levels. In a single participant, further
In the field of endurance sports, the nutritional approaches development of the product resulted in even greater increases
used to optimally fuel performance are undergoing a in BHB with no change in blood glucose.
paradigm shift. In dramatic opposition to the high-
carbohydrate diets and frequent ‘carb-loading’ strategies During prolonged endurance exercise, especially events of
previously used by endurance athletes, many elite long- moderate intensity (i.e. up to ≈50% of VO2Max), aerobic fat
distance runners, triathletes and cyclists are now employing metabolism provides at least 50% of total energy expenditure
regular carbohydrate restriction, including during races, to (TEE).24 The energy contribution from fat oxidation then
improve performance.8,20,21 This approach was largely decreases dramatically as intensity increases above 60% of
pioneered by the work of Jeff Volek and Stephen Phinney, as VO2Max. However, extended periods of low-carbohydrate

http://www.insulinresistance.org Open Access

Page 6 of 9 Original Research

120 PFv1 a during the recovery period.21 Other than the signalling effects
PFv2 of lower carbohydrate availability during exercise, the anti-
inflammatory and epigenetic (via histone deacetylase
110
inhibition) effects of elevated ketone bodies such as BHB
after carbohydrate restriction may provide other beneficial
Glucose (mg/dl)

100 adaptations for the endurance athlete.27,28

90 Despite a greater contribution of fatty acids to TEE during

aerobic exercise, keto-adapted athletes deplete muscle
80 glycogen at a similar rate to athletes consuming a
carbohydrate-based diet.8 This suggests that a carbohydrate
70
source is still required during prolonged exercise in these
0 1 2 3 4 athletes. In this study, a 25 g dose of PF resulted in a small
Time (hours) (median 6 mg/dl) increase in blood glucose within 30 min.
The carbohydrate source in PF is therefore not entirely
1.0 PFv1 b
PFv2 digestion resistant. However, the glucose response at 30 min
was highly variable between participants, ranging from -2.8
Blood ketones (mmol/L)

0.8
mg/dl to +16.0 mg/dl. This may be due to differences in
glycogen levels and insulin sensitivity after an overnight fast,
0.6 or occur as a result of differences in genetics and the gut
microbiota affecting metabolic responses to carbohydrate.29
0.4 Additionally, one participant (#3) reported drinking black
coffee during the second hour of the study. Caffeinated coffee
0.2
has been shown to acutely increase blood glucose,30 and this
participant also had the highest blood glucose levels at both
2 h (106 mg/dl) and 3 h (103 mg/dl).
0.0
0 1 2 3 4
Time (hours) Although PF was originally intended to not include a
significant source of glucose, a small glycaemic load may
Source: Authors’ own work
actually be of benefit to the low-carb endurance athlete if
FIGURE 7: Glucose and BHB responses to PFv2. (a) In participant #4, who had
displayed the greatest blood glucose response to PFv1, 25 g of PFv2 did not provided before or during exercise. One major use of glucose
elevate blood glucose from a baseline of 88 mg/dl. (b) A greater BHB response during exercise in athletes consuming a low-carbohydrate
was seen with PFv2 compared to PFv1. After 25 g of PFv2, BHB increased by 0.5
mmol/L from 0.2 mmol/L at baseline to 0.7 mmol/L after 30 min, reached a peak diet is the regeneration of oxaloacetate (OAA) in the
of 0.8 mmol/L at 2 h and remained above 0.5 mmol/L for the whole study mitochondrial Kreb’s cycle, a process known as anapleurosis.
period.
This occurs when mitochondrial acetyl-CoA accumulates,
which increases the activity of pyruvate carboxylase.
keto-adaptation in endurance athletes have recently been
Pyruvate carboxylase diverts pyruvate from glycolysis to the
shown to result in the capacity to perform peak fat oxidation
production of OAA that can bind with acetyl-CoA to form
at over 80% of VO2Max.8 In the same study, aerobic fat
citrate as the ‘first step’ of the Kreb’s cycle. In the low-carb
metabolism provided an average 88% of TEE during a 180-
athlete, mitochondrial acetyl-CoA may accumulate in the
min treadmill run at 65% of VO2Max in a group of 10 elite skeletal muscle due to increased lipolysis and beta-oxidation
ultra-marathoners and ironman distance triathletes of fatty acids, or as the result of increased ketone availability
habituated on a low-carbohydrate diet (< 10% of calories (transesterification and cleavage of acetoacetate to two
from carbohydrate).8 The keto-adapted athletes displayed molecules of acetyl-CoA). To help drive aerobic production
greater ketone and glycerol levels during prolonged exercise of ATP, this acetyl-CoA must then bind to OAA. Adequate
compared to athletes eating a carbohydrate-based diet, production of OAA is therefore a rate-limiting aspect of the
despite similar circulating levels of insulin and glucose.8 ability to use fatty acids and ketone bodies as fuel during
The increase in capacity to metabolise fat at higher intensities exercise. This is thought to be the reason why exogenous
is therefore likely to be due to metabolic changes such as ketones (given as a ketone ester) have a greater performance-
upregulation of pathways involved in aerobic fatty acid boosting effect when given with a source of glucose (Clarke
metabolism, as well as selective changes in tissue insulin K 2016, personal communication, October 13).13
sensitivity. Restricting carbohydrates may also result in
greater glycogen depletion, increasing AMPK activation and As fat-adapted athletes do not appear to spare glycogen
promoting pathways associated with mitochondrial during exercise despite the majority of TEE coming from
biogenesis.21,25 However, other data suggest that endurance fatty acids, and exogenous ketone sources have a greater
exercise itself is the largest stimulus for mitochondrial benefit when given with a source of carbohydrate, glucose
biogenesis regardless of carbohydrate timing26 and that the availability in the muscle (either from glycogen stores or
signalling benefits of training in a carbohydrate-restricted from an exogenous source) still has the potential to be a rate-
state may be enhanced by consuming a carbohydrate source limiting factor to performance in the low-carb athlete.

http://www.insulinresistance.org Open Access

Page 7 of 9 Original Research

One other product, a hydrothermally processed corn starch results in greater ketone production and a smaller glycaemic
(UCAN superstarch), has been specifically marketed to effect would be beneficial. This was the reason behind the
athletes eating a low-carbohydrate diet for this reason.9 As development of PFv2. In separate trials by one participant,
well as providing a low glycaemic index carbohydrate reducing the digestion-resistant maltodextrin content by 40%
source, similar to that used in UCAN, one potential benefit of and changing the MCT source to predominantly caprylic
PF is that that majority of calories come from MCTs, which acid (C8:0) doubled the maximal increase in BHB without
can boost endogenous ketone levels. In this manner, PF affecting blood glucose. This suggests that PFv2 would also
provides an anapleurotic substrate and ketone source be suitable as an MCT powder to support ketogenic diets
together. However, as with the blood glucose response, BHB where maintenance of lower blood glucose levels is
responses to PF were highly variable, ranging from -0.2 important, such as for certain cancers or neurological
mmol/L to +0.6 mmol/L after 30 min. Excluding the results diseases.34 MCT-based products may also be useful in
from one outlier (participant #8) who had a fasting BHB of 1.1 supporting the loss of excess adipose tissue, as MCTs have
mmol/L, PF resulted in a significant increase in BHB, with a been shown to both decrease appetite and increase diet-
median response of +0.2 mmol/L. As 25 g of PF contains induced thermogenesis.35,36,37 One significant benefit of the PF
around 15 g of MCTs, this increase in BHB is about as much formulation is that no participant reported any negative GI
as would be predicted.3 The data also suggest that there is a symptoms, which are the most frequently reported side
cut-off level of BHB somewhere between 0.5 mmol/L and 1.0 effects of MCT consumption.3 This supports the idea that
mmol/L where consuming PF will either increase BHB (due MCTs are better tolerated when delivered as a powder.
to provision of MCTs as a source of ketones) or decrease BHB
(due to the provision of an anapleurotic substrate). However, This study does have some limitations. Firstly, this was a
other factors such as insulin sensitivity and glucose series of open-label pilot studies, and therefore, the
transporter activity will also be involved, as the individual participants were not blinded. While it is likely that the acute
BHB response was not correlated with concurrent changes in
metabolic effects were due to PF, the nature of the study may
glucose or insulin levels. The dynamics of the BHB response
have prevented the participants from openly reporting
to PF agrees with data from other studies, which show a
negative subjective side effects. Although generally
biphasic response after consuming MCTs. For instance, portal
considered to be the standard measure of blood ketone levels,
vein medium-chain fatty acid levels peak at 15 min and 75–90
using blood BHB to determine the ketogenic capacity of an
min after infusion of MCTs into the GI tract of pigs,31 and
intervention gives an incomplete picture. Absolute BHB
blood BHB levels peak around 30 min and 3 h after MCT
levels are a dynamic sum of ketone production (in the liver),
consumption in humans.32 This is the same pattern seen in
usage (in the peripheral tissues) and wastage (i.e. in the urine
this study. However, the BHB response to MCT consumption
and breath). Other factors such as the hepatic NAD+/NADH
also appears to be greater after a period of regular MCT
ratio will determine the relative abundances of BHB and
consumption.31 Different degrees of habitual MCT
acetoacetate (AcAc, which can also be decarboxylated to
consumption may therefore account for some of the
acetone),38 the latter of which is less routinely measured in
variability in BHB responses between participants. Although
the blood outside of the laboratory setting. Importantly, as
the early (30 min) BHB peak is likely to be solely due to PF,
the late (3–4 h) increase in BHB may also in part be due to a support increases for the use of performance-enhancing
prolonged period of fasting overnight followed by consuming supplements that increase ketone levels exogenously,39 it is
only around 150 kcal of energy, most of which was fat. worth noting that the pure elevation of blood BHB levels is
not the goal of supplementation – any additional ketone
To provide an optimal fuel source for low-carb or keto- bodies must also be readily available as an efficient metabolic
adapted athletes during exercise, any supplement should substrate. In the case of supplements such as PF, the goal is to
have minimal effects on insulin levels, as endogenous provide a readily oxidisable calorie source that does not
lipolysis must still be allowed to contribute the majority of negatively affect the athlete’s underlying physiology. As
TEE. Consuming PF resulted in a small but non-significant ketones are actively transported into the exercising muscle to
increase in insulin after 30 min, with the response ranging support aerobic work,11 and the enzymes of ketolysis are
from -2.5 µU/ml to +11.0 µU/ml. The degree of insulin upregulated by endurance exercise,39 it is likely that PF can
response appeared to be associated with fasting insulin level, provide a calorie source to support longer endurance efforts
with minimal responses seen in those with a fasting insulin in the low-carb athlete without ever causing unnecessary
level above 5 µU/ml (but below 10 µU/ml). Aside from one deviations in blood glucose or BHB.
participant (#1), insulin remained below 10 µU/ml
throughout. This is the approximate level at which lipolysis Another potential limitation is the fact that a linear regression
is thought to be maximally suppressed in healthy had to be constructed to adjust the reported blood glucose
individuals.15,33 However, it is worth considering that glucose results, as the glucose levels from the blood spots were
and insulin levels both decrease during endurance exercise, spuriously low. A similar method has previously been used to
even when a high glycaemic index carbohydrate source is reduce the error in the delayed measurement of blood glucose
administered.9 Therefore, both insulin and glucose responses from filter paper,40 though there is an inherent error into this
to PF would be smaller if given during exercise. Despite this, approach, as the calibration was based on 18 data points from
there are a number of scenarios where a supplement that only two individuals. A number of studies have shown that

http://www.insulinresistance.org Open Access

Page 8 of 9 Original Research

blood glucose levels on filter paper decrease over time and designed the study protocol, performed the statistical
with increasing temperature, though they are generally analysis, and drafted and revised the manuscript.
thought to remain stable for up to seven days at room
temperature.41,42 Time and conditions could therefore have References
promoted glucose degradation or increased erythrocyte
1. Volek JS, Noakes T, Phinney SD. Rethinking fat as a fuel for endurance exercise. Eur
metabolism of glucose in situ during transport of the blood J Sport Sci. 2015;15:13–20. https://doi.org/10.1080/17461391.2014.959564
spots back to the laboratory. Importantly, as paired statistical 2. Taubes G. The science of obesity: What do we really know about what makes us fat?
An essay by Gary Taubes. BMJ. 2013;346:f1050. https://doi.org/10.1136/bmj.f1050
analyses were performed, all blood spots that were directly
3. Cunnane SC, Courchesne-Loyer A, Vandenberghe C, et al. Can ketones help rescue
compared (i.e. from within the same participant) were exposed brain fuel supply in later life? Implications for cognitive health during aging and
to the same transport conditions before analysis. Similarly, a the treatment of Alzheimer’s disease. Front Mol Neurosci. 2016;9:53. https://doi.
org/10.3389/fnmol.2016.00053
number of insulin results had to be assigned an insulin result 4. Storoni M, Plant GT. The therapeutic potential of the ketogenic diet in treating
due to the level being below the detection limit of the assay. progressive multiple sclerosis. Mult Scler Int. 2015;2015:681289. https://doi.
org/10.1155/2015/681289
However, non-parametric analysis of the data means that the 5. Paoli A. Ketogenic diet for obesity: Friend or foe? Int J Environ Res Public Health.
absolute glucose and insulin levels do not influence the results, 2014;11:2092–2107. https://doi.org/10.3390/ijerph110202092
as the analyses are based on ranks. Although the number of 6. Feinman RD, Pogozelski WK, Astrup A, et al. Dietary carbohydrate restriction as
the first approach in diabetes management: Critical review and evidence base.
participants included is fairly standard for a study of this type, Nutrition. 2015;31:1–13. https://doi.org/10.1016/j.nut.2014.06.011
it is also possible that changes in insulin over the first hour of 7. Marquet L-A, Brisswalter J, Louis J, et al. Enhanced endurance performance by
periodization of carbohydrate intake: ‘Sleep Low’ strategy. Med Sci Sports Exerc.
the study may have been significant if more participants were 2016;48:663–672. https://doi.org/10.1249/MSS.0000000000000823
included. Finally, although the issues regarding participant- 8. Volek JS, Freidenreich DJ, Saenz C, et al. Metabolic characteristics of keto-adapted
ultra-endurance runners. Metabolism. 2016;65:100–110. https://doi.org/10.​
led remote collection of samples resulted in some potential 1016/j.metabol.2015.10.028
errors within the data, a strength of the study is that it was 9. Roberts MD, Lockwood C, Dalbo VJ, et al. Ingestion of a high-molecular-weight
hydrothermally modified waxy maize starch alters metabolic responses to
carried out in the participants’ own homes rather than in a prolonged exercise in trained cyclists. Nutrition. 2011;27:659–665. https://doi.
highly controlled laboratory setting. The study therefore org/10.1016/j.nut.2010.07.008
10. Bach AC, Babayan VK. Medium-chain triglycerides: An update. Am J Clin Nutr.
captures the inherent variability that all people will experience 1982;36:950–962.
when using supplements in the real world, which will require 11. Balasse EO, Féry F. Ketone body production and disposal: Effects of fasting,
a degree of self-experimentation and monitoring to ensure diabetes, and exercise. Diabetes Metab Rev. 1989;5:247–270. https://doi.
org/10.1002/dmr.5610050304
optimal fuelling for individual performance. 12. Sato K, Kashiwaya Y, Keon CA, et al. Insulin, ketone bodies, and mitochondrial
energy transduction. FASEB J. 1995;9:651–658.

In summary, we show that 25 g of an MCT oil-based sports 13. Cox PJ, Kirk T, Ashmore T, et al. Nutritional ketosis alters fuel preference and
thereby endurance performance in athletes. Cell Metab. 2016;24:256–268.
supplement powder increases average BHB levels by 0.2 https://doi.org/10.1016/j.cmet.2016.07.010
mmol/L with only a small increase in blood glucose and 14. Furuyashiki T, Tanimoto H, Yokoyama Y, et al. Effects of ingesting highly branched
cyclic dextrin during endurance exercise on rating of perceived exertion and blood
without significantly increasing insulin levels. A revised components associated with energy metabolism. Biosci Biotechnol Biochem.
2014;78:2117–2119. https://doi.org/10.1080/09168451.2014.943654
formulation of the supplement resulted in twice the BHB
15. Jensen MD, Caruso M, Heiling V, et al. Insulin regulation of lipolysis in nondiabetic
response without affecting blood glucose. For the low-carb or and IDDM subjects. Diabetes. 1989;38:1595–1601. https://doi.org/10.2337/diab.​
38.12.1595
keto-adapted athlete, both PF and PFv2 have the potential to
16. Chassaing B, Koren O, Goodrich JK, et al. Dietary emulsifiers impact the mouse gut
support both ketone production and anapleurosis in order to microbiota promoting colitis and metabolic syndrome. Nature. 2015;519:92–96.
https://doi.org/10.1038/nature14232
help promote optimal substrate utilisation during exercise.
17. Lerner A, Matthias T. Changes in intestinal tight junction permeability associated
Future studies will provide a larger dose of PFv2 and with industrial food additives explain the rising incidence of autoimmune
determine the metabolic responses during exercise compared disease. Autoimmun Rev. 2015;14:479–489. https://doi.org/10.1016/j.autrev.​
2015.01.009
to a placebo. 18. Chassaing B, Gewirtz AT. Has provoking microbiota aggression driven the obesity
epidemic? BioEssays. 2016;38:122–128. https://doi.org/10.1002/bies.201500116

Acknowledgements 19. Sicherer SH, Sampson HA. Food allergy: Epidemiology, pathogenesis, diagnosis,
and treatment. J Allergy Clin Immunol. 2014;133:291–307; quiz 8. https://doi.
org/10.1016/j.jaci.2013.11.020
The authors would like to thank all the participants for their 20. Maffetone PB, Laursen PB. Reductions in training load and dietary carbohydrates
time and input to the study. They would also like to thank help restore health and improve performance in an Ironman triathlete. Int J
Sports Sci Coach. 2017; Online publication ahead of print.
Ms. Amelia Luker for her assistance assembling study kits 21. Impey SG, Hammond KM, Shepherd SO, et al. Fuel for the work required: A
and collecting the study data, as well as Dr Elizabeth Nance practical approach to amalgamating train-low paradigms for endurance athletes.
Physiol Rep. 2016;4. pii: e12803. https://doi.org/10.14814/phy2.12803
for her input during the preparation of the manuscript.
22. Noakes T, Volek JS, Phinney SD. Low-carbohydrate diets for athletes: What
evidence? Br J Sports Med. 2014;48:1077–1078. https://doi.org/10.1136/
bjsports-2014-093824
Competing interests 23. Maffetone PB, Laursen PB. Athletes: Fit but unhealthy? Sports Med Open.
2015;2:24. https://doi.org/10.1186/s40798-016-0048-x
Both T.R.W. and C.K. are employees of Nourish Balance 24. Egan B, Zierath JR. Exercise metabolism and the molecular regulation of skeletal
Thrive and co-creators of PF and PFv2. All profits to date muscle adaptation. Cell Metab. 2013;17:162–184. https://doi.org/10.1016/j.
cmet.2012.12.012
from the sales of PF were used to fund the current study. 25. Bartlett JD, Louhelainen J, Iqbal Z, et al. Reduced carbohydrate availability
enhances exercise-induced p53 signaling in human skeletal muscle: Implications
for mitochondrial biogenesis. Am J Physiol Regul Integr Comp Physiol.
Authors’ contributions 2013;304:R450–R458. https://doi.org/10.1152/ajpregu.00498.2012
26. Jensen L, Gejl KD, Ørtenblad N, et al. Carbohydrate restricted recovery from long
C.K. recruited study participants, provided study materials term endurance exercise does not affect gene responses involved in mitochondrial
biogenesis in highly trained athletes. Physiol Rep. 2015;3. pii: e12184. https://doi.
and assisted in the preparation of the manuscript. T.R.W. org/10.14814/phy2.12184

http://www.insulinresistance.org Open Access

Page 9 of 9 Original Research

27. Shimazu T, Hirschey MD, Newman J, et al. Suppression of oxidative stress by beta- 35. Van Wymelbeke V, Himaya A, Louis-Sylvestre J, et al. Influence of medium-chain
hydroxybutyrate, an endogenous histone deacetylase inhibitor. Science. and long-chain triacylglycerols on the control of food intake in men. Am J Clin
2013;339:211–214. https://doi.org/10.1126/science.1227166 Nutr. 1998;68:226–234.
28. Youm YH, Nguyen KY, Grant RW, et al. The ketone metabolite beta-hydroxybutyrate 36. St-Onge MP, Mayrsohn B, O’Keeffe M, et al. Impact of medium and long chain
blocks NLRP3 inflammasome-mediated inflammatory disease. Nat Med. triglycerides consumption on appetite and food intake in overweight men. Eur J
2015;21:263–269. Clin Nutr. 2014;68:1134–1140. https://doi.org/10.1038/ejcn.2014.145
29. Zeevi D, Korem T, Zmora N, et al. Personalized nutrition by prediction of glycemic 37. Hill JO, Peters JC, Yang D, et al. Thermogenesis in humans during overfeeding with
responses. Cell. 2015;163:1079–1094. https://doi.org/10.1016/j.cell.2015.11.001 medium-chain triglycerides. Metabolism. 1989;38:641–648. https://doi.
org/10.1016/0026-0495(89)90101-7
30. Robertson TM, Clifford MN, Penson S, et al. A single serving of caffeinated coffee
impairs postprandial glucose metabolism in overweight men. Br J Nutr. 38. Krebs HA, Gascoyne T. The redox state of the nicotinamide-adenine dinucleotides
2015;114:1218–1225. https://doi.org/10.1017/S00071145​15002640 in rat liver homogenates. Biochem J. 1968;108:513–520. https://doi.org/10.1042/
bj1080513
31. Guillot E, Vaugelade P, Lemarchal P, et al. Intestinal absorption and liver uptake of
medium-chain fatty acids in non-anaesthetized pigs. Br J Nutr. 1993;69(2):​431– 39. Evans M, Cogan KE, Egan B. Metabolism of ketone bodies during exercise and
442. training: Physiological basis for exogenous supplementation. J Physiol. 2016.
https://doi.org/10.1113/JP273185
32. Courchesne-Loyer A, Fortier M, Tremblay-Mercier J, et al. Stimulation of mild,
sustained ketonemia by medium-chain triacylglycerols in healthy humans: 40. Cox DJ, Moll ME, Gonder-Frederick LA, et al. Accuracy of time-delayed filter-paper
Estimated potential contribution to brain energy metabolism. Nutrition. blood glucose determination. Diabetes Care. 1988;11:517–518. https://doi.
2013;29(4):635–640. https://doi.org/10.1016/j.nut.2012.09.009 org/10.2337/diacare.11.6.517
33. Sonksen P, Sonksen J. Insulin: Understanding its action in health and disease. Br J 41. Abyholm AS. Determination of glucose in dried filter paper blood spots. Scand J
Anaesth. 2000;85:69–79. https://doi.org/10.1093/bja/85.1.69 Clin Lab Invest. 1981;41:269–274. https://doi.org/10.3109/00365518109092044
34. Meidenbauer JJ, Mukherjee P, Seyfried TN. The glucose ketone index calculator: A 42. Winocour PH, McKinnon GA, McMurray JR, et al. Evaluation of the
simple tool to monitor therapeutic efficacy for metabolic management of brain measurement of blood glucose levels on dried filter paper blood spots. Diabet
cancer. Nutr Metab. 2015;12:12. https://doi.org/10.1186/s12986-015-0009-2 Med. 1985;2:269–271.

http://www.insulinresistance.org Open Access