ERD

Examine.com
Research Digest

Issue 4  ◆  February 2015 1

Table of Contents
05 Mood, dieting, and macros

14 What If There Were No Dietary Guidelines?

17 The iPad Hangover

25 Can mice get cancer from steak?

33 Sodium phosphate: a potentially underutilized

    ergogenic aid

39 On the whey to getting lean: one more round of

    whey vs. soy

45 It’s (not) all in your head: how sodium intake

    affects headaches

51 Diets, fast and slow

59 Is the glycemic index actually useful for

    making food choices?

66 INTERVIEW: Ivan Oransky

69 INTERVIEW: Jessica Richman

2
From the Editor
Let’s play a little game of Choose Your Own Adventure,
Health Research Edition. It’ll only take a few minutes, but I
hope it gets you thinking.
You’re going to be playing the role of someone scouring the
web for evidence on a condition you’ve just been diagnosed
with. The title is “The Case of the Misleading Abstract.”
Foreshadowing? Yes.
On a random weeknight, you’re supposed to be sleeping
but are instead doing health research (hey, it should pay off
eventually, right?). Since you haven’t read the February issue
of ERD, you don’t yet know how nighttime web-browsing is
affecting your brain. Anyway, you stumble across an inter-
esting abstract on pubmed. Apparently, alpha lipoic acid can
help treat symptoms of a certain autoimmune disease. Not
only is p less than 0.05, it’s less than freaking 0.001! Boom.
This is quite the useful bit of knowledge, because you’ve
just been diagnosed with that autoimmune disease, and
your doctor wants to put you on an unpleasant medication.
Should you (A) go buy the supplement on Amazon, or (B)
try to pull the full paper?
(A) You buy the alpha lipoic acid on Amazon
It arrives in two days. You don’t mention this to your doctor
because he’s a bit close-minded, but you’re sort-of banking
on the supplement helping, and hence delay starting your
medication for a few weeks. A few weeks later, the pain hasn’t
gone away, you’re still fatigued, and you have a new-found
hate for alpha lipoic acid.
Kamal Patel, Editor-in-Chief
(B) You pull the full paper
You can’t actually access the full paper. Your significant
other is a student, so eventually you pull the full paper and
get all the details. The researchers used three different ques-
tionnaires and took some blood draws, and out of all those
measurements, only one was significantly better in the
alpha lipoic acid group. You look up one of the references
mentioned in the paper, and it actually concludes that alpha
lipoic acid isn’t likely to impact autoimmunity by itself.
Also, the most convincing studies have been conducted on
diabetic mice.
You don’t buy the alpha lipoic acid. You have a little bit of
“nonbuyer’s remorse,” but forget about it the next week. The
$25 you saved becomes $436 by the time you retire.
Neither of these scenarios is uncommon. Most of us have
bought way too many supplements based on either anec-
dotes or tantalizing p-values. Granted, it’s really tough to
face a health condition without an easy treatment, and that
can make an abstract sound more promising than the full
paper or body of research reveals. Money spent on ineffec-
tive supplements can really add up, and sometimes it may
be a better idea to focus on high-yield lifestyle habits like
plenty of sleep and methods to combat stress. Don’t take
abstracts and their p-values as gospel, and best of luck on
choosing your own adventure.
3
Contributors
Researchers

Trevor Kashey Alex Leaf Courtney Silverthorn Zach Bohannan Anders Nedergaard Jeff Rothschild
Ph.D(c) M.S(c) Ph.D. M.S. Ph.D. M.Sc., RD

Editors

Gregory Lopez Pablo Sanchez Soria Kamal Patel

Pharm.D. Ph.D. M.B.A., M.P.H.,
Ph.D(c)

Reviewers

Arya Sharma Natalie Muth Stephan Guyenet Sarah Ballantyne Katherine Rizzone Spencer Nadolsky
Ph.D., M.D. M.D., M.P.H., RD Ph.D. Ph.D. M.D. D.O.

Mark Kern Gillian Mandich

Ph.D., RD Ph.D(c)

4
Mood, dieting, and macros
Transient decrements in mood during
energy deficit are independent of dietary
protein-to-carbohydrate

Introduction mass during energy restriction in sedentary and athletic pop-

ulations. These benefits are important for dietary adherence
Weight loss is big business. Some authorities estimate that 45
and long-term success, but they are only a piece of the puzzle.
million Americans are trying to lose weight each year and
spend upwards of $33 billion annually to do so. And it has
Dieting can be psychologically complex. The brain and ner-
been suggested that overweight and obese people experience
vous system communicate through small chemicals called
significant improvements on a range of subjective symptoms
neurotransmitters. Collectively, these neurotransmitters are
after weight loss, regardless of diet composition.
what allow us to be aware, have emotion, remember things,
move our body, regulate body temperature, sleep, and do or
However, dieting is also common in healthy-weight peo-
feel anything that our brain allows for. In fact, many dis-
ple looking to improve body composition and/or athletic
eases like Parkinson's, Alzheimer’s, depression, insomnia,
performance. Consuming high-protein diets has become a
ADHD, and anxiety have been linked to neurotransmitter
popular method to aid in weight loss, as research has shown
imbalances.
high-protein diets suppress hunger and preserve lean body

5
 Figure 1: Select roles of acids (BCAAs) – have been shown to alter brain neuro-
chemistry through basic competition. That is, these amino
mood-related neurotransmitters
acids all share the same transporters that allow access into
the brain, and thus they all compete for entry. Since the
transporters are not specific to any one of the amino acids,
the largest determinant of which enters the brain is their
concentrations. Thus, if the plasma level of the BCAAs
increases, then brain concentrations of tryptophan, tyro-
sine, and their respective neurotransmitters is reduced.
Theoretically, this may have negative consequences for
mood, sleep, hunger, and overall liveliness.

On the other hand, carbohydrate intake has been observed

to increase serotonin production secondary to insulin pro-
moting tryptophan uptake in the brain. Theoretically, this
may benefit mood. > The study under review aimed to com-
pare the effects of different dietary protein-to-carbohydrate
ratios on cognitive performance, mood, and sleep quality
during short-term energy restriction.

Who and what was studied?

The outcomes presented in this study were actually second-

There are many ways to classify neurotransmitters, but for ary outcomes that were collected in a previous controlled

our purposes it’s especially important to understand the trial. That trial aimed to evaluate the effects of protein

role that amino acids play in different neurotransmitters. intake on body composition, protein balance, and calcium

Some amino acids, such as glycine, taurine, and glutamate, homeostasis during a short-term energy deficit.

serve directly as neurotransmitters, whereas other amino

acids like tyrosine and tryptophan serve as precursors for Thirty-nine volunteers (82% men) with an average age of 21

neurotransmitter synthesis. Tyrosine is the precursor for years and average BMI of 25 completed the study. Despite the

the synthesis of the catecholamines: dopamine, adrenaline technically “overweight” BMI, the inclusion criteria required

(epinephrine), and noradrenaline (norepinephrine). These study participants to be recreationally active (defined as

neurotransmitters play a central role in attention, learning, three to four days a week of aerobic and/or resistance exer-

motivation, and alertness. Tryptophan serves as the pre- cise) and physically fit (VO2peak of 40–60 mL/kg/min).

cursor for serotonin, which can have an indirect effect on

well-being and happiness, and plays a variety of other roles The volunteers spent 31 days in a metabolic ward under

as well. constant supervision to ensure compliance to the prescribed

diets and physical activity. Using a randomized block

There is some controversy surrounding high-protein diets design, they were divided into three groups differing in pro-

because the consumption of a lot of large neutral amino tein and carbohydrate content, but matched for total caloric

acids – tyrosine, tryptophan, and the branch-chained amino intake and fat:

6
 • Low-protein (LP) group consuming 0.8 g/kg/day protein
• Moderate-protein (MP) group consuming 1.6 g/kg/day
Stratified block
• High-protein (HP) group consuming 2.4g/kg/day
randomization
The increase in protein was accommodated by an equal reduction in
carbohydrates. Throughout the entire 31-day study period, protein was Randomization methods are very
held constant at the assigned amount, fat was roughly 30% of total energy important for avoiding potential
intake, and carbohydrates made up the remainder. Alcohol and smok- sources of bias, yet often are not
ing were forbidden, and all the subjects had “lights out” by 11:00 p.m. to reported adequately in journal articles.
ensure similar and adequate sleep between the groups. Nutritional sup- When it comes to designing clinical tri-
plements were also forbidden, with the exception of a daily multivitamin al groups, balance is key. If two groups
provided by the researchers. are tested, and one of the groups is
a good deal healthier than the other
The first ten days of the study served as the control period, with subjects one, the results are likely to be biased.
practicing weight maintenance. Their energy expenditure was estimated
from pre-study indirect calorimetry measurements and physical activity Randomization solves much of this
logs. During the subsequent 21 days, energy intake was reduced by 30% problem, by giving each subject an
(through reductions in fat and carbohydrates) and physical activity energy equal likelihood of being in different
expenditure was increased by 10%, for a total daily deficit of 40% com- study groups. But what happens when
pared to weight maintenance. the sample size is small (as it often is
in trials, as huge intervention studies
All diets were prepared by research dietitians and protein was provided as require huge amounts of money)?
mixed, high-quality proteins (e.g. dairy, lean meats, and vegetable-based The smaller the sample size, the more
proteins). The subjects all performed regular physical activity. This included likely it is that groups will differ in char-
three days a week of low-volume resistance training, where the participants acteristics that affect the outcome.
performed one single-joint movement per major muscle group (three sets You can statistically control for known
of 15 repetitions) using workloads determined during the pre-study period, factors, but you can’t control for fac-
plus daily steady-state endurance activity at a low- to moderate-intensity tors that aren’t measurable or that you
(40–60% VO2peak) pace on a treadmill or stationary bicycle. don’t know about.

A battery of cognitive tests were administered on days 11, 20, and 30 in As opposed to simple randomization,
the afternoon, about four hours after lunch. These time points were select- randomization through a stratified
ed to represent the beginning (day 11), the midpoint (day 20), and end block design allows the researchers to
(day 30) of the dieting period. During the control period, three practice statistically “remove” variables of the
sessions were allowed for familiarization and to help reduce the potential study subjects that could potentially
of a learning effect (performing better because of practice rather than influence the study results but aren’t
the variable of interest). During the cognitive testing, self-reported mood being studied. When randomly allocat-
state was also assessed, utilizing the Profile of Mood States questionnaire. ing the subjects into the experimental
This questionnaire asks the question, “how are you feeling right now?” and control groups, the researchers
and has the responder rate 65 mood-related adjectives on a zero to four first stratify them into “blocks” based
point scale. on traits they wish to hold constant

7
Self-reported sleep quality was assessed every morning
upon waking using a questionnaire asking how long it
took to fall asleep, alertness at bedtime and upon waking,
number of awakenings during the night, restedness upon
waking, and four questions used to calculate an overall
index of sleep quality.
Finally, blood samples were collected following an overnight
fast on days 10 and 31 to quantify plasma amino acid levels.
In this study, 39 people were kept under observa-
tion for a month while progressively consuming
fewer calories and engaging in more physical
activity as the study went on. Researchers adminis-
tered cognitive tests and measured plasma amino
acid levels.
What were the findings?
The original trial that provided data for this study assessed
the impact of protein intake on body composition. All three
groups lost weight during the energy deficit, but the total fat
loss as a percentage of weight loss was greater and the total
fat-free mass loss smaller in the moderate- and high-protein
Figure 2: Protein & carb intake by study period
groups compared to the low-protein group. The diets during
both the energy balance (first 10 days) and energy deficit
phases can be seen in Figure 2.
Total mood disturbances, anger, and tension significant-
ly increased in all groups from day 11 to day 20, but were
not significantly different from day 11 to day 30 and were
not significantly different between any of the groups.
Additionally, confusion and vigor tended to increase during
the energy deficit with no difference between groups.
Depression and fatigue were the only mood-states that did
not differ throughout the intervention.
There was also a lack of differences between any groups on
the cognitive tests throughout the intervention. Although
there were nine cognitive tests administered, only one test
(the Four-Choice Reaction Time test) differed in any of the
group comparisons, with the high protein group perform-
ing better than the moderate protein group.
While all groups tended to improve on cognitive tests during
the energy deficit period, the improvements were observed in
tests that also happened to improve during every practice ses-
sion within the energy balance phase. This suggests that the
improvements through the energy deficit period were a result
of a learning effect rather than being due to the energy deficit.
8
There were also no significant differences in sleep quality
throughout the intervention in any group or between the
groups, with the one exception being a significant reduction
in alertness upon waking in the low-protein group from day
11 to day 20, which returned to normal at day 30.
Finally, plasma BCAA concentrations were significant-
ly higher and tyrosine concentrations significantly lower
in the moderate- and high-protein groups relative to the
low-protein group on day 11, but the difference between
the high- and moderate-protein groups was not significant.
Moreover, upon entering energy deficit (days 20 and 30), all
groups showed significant elevations in plasma BCAA and
reductions in tyrosine concentrations, with the between-
group differences losing their significance. However there
  Total mood
disturbances,
anger, and
tension [...] were
not significantly
different between
any of the groups,
suggesting that
diet composition
had no effect on
mood changes.
was a slight trend for the high- and moderate-protein groups
to retain their difference relative to the low-protein group.
The biggest shift in mood happened after day 11.
Study participants felt significantly more anger
and tension, though depression and fatigue were
not affected. Mood did not change much after
day 20. The results were fairly consistent between
groups eating different amounts of protein,
though the high- and moderate-protein groups
had higher BCAA concentrations and lower tyro-
sine levels than the low-protein group.
What does the study really
tell us?
The purpose of this study was to examine how the dietary
protein-to-carbohydrate ratio influenced cognitive perfor-
mance, mood, and subjective sleep quality. The premise is
that manipulating dietary protein and carbohydrate intake
would modulate neurotransmitter synthesis (primarily
serotonin and dopamine). It was indeed shown that plasma
BCAA and tyrosine, but not tryptophan, concentrations
were impacted by the dietary treatments in a relatively short
period of time, as the differences were noticeable by the end
of the ten-day control period. However, at this time there
were no differences between the groups in cognition, mood,
or sleep.
At the end of the dieting period, the differences in plasma
BCAA and tyrosine concentrations became insignificant.
However, there was a trend towards significance, suggesting
that a larger sample size or longer duration may have been
needed to detect the differences. Regardless, mood changes
were not different between the groups and were transient,
returning to pre-dieting levels by the end of the interven-
tion. This suggests that it may have been the caloric deficit,
not the plasma amino acid concentrations, that increased
anger, tension, and total mood disturbances.
9
Similar things can be said for cognition, with all but one
change occurring without difference between the groups.
Finally, subjective sleep was mostly unaffected during the
entire study. The one exception was the low-protein group
feeling less alert upon waking during the first ten days of
the dieting period only. Seeing how dopamine and nor-
adrenaline play a central role in alertness, this outcome
could be the result of the reduction of plasma tyrosine con-
centrations when the dieting period began. However, this
would not explain why the sleep effects were not seen in the
other groups.
So overall, what does this study tells us? It shows that a 40%
caloric deficit in lean and recreationally active young adults
[...] boosting protein probably won’t help
you avoid the doldrums of starting a diet.
The initial ten days of beginning a diet
may be the most psychologically difficult,
but that this will pass with time.
may have a transient effect on mood regardless of the pro-
tein-to-carbohydrate ratio. It also tells us that the dietary
protein-to-carbohydrate ratio does impact plasma concen-
trations of BCAAs and tyrosine, but not tryptophan. This
could be owed in part to the dietary supply of amino acids,
with all groups consuming 16 times more BCAAs and seven
times more tyrosine and phenylalanine than tryptophan.
Nonetheless, this suggests that downstream effects of the
dietary protein to carbohydrate ratio would be limited to
the catecholamines and not serotonin.
However, tryptophan entry into the brain is facilitated by
insulin, which acts to reduce plasma BCAA and tyrosine
levels by shuttling them into skeletal muscle and organs,
and consequently also reduces competition for entry into
the brain. Blood amino acid measurements were done after
an overnight fast, when insulin is minimal. Similarly, the
cognitive tests and mood questionnaire were completed
roughly four hours after lunch, when insulin levels would
be expected to have returned to baseline. Thus, it is possible
that different outcomes would have been seen if the tests
were administered closer to lunch when the subjects were in
the postprandial state.
Another limitation of the study is that it was originally
powered to detect changes in muscle protein synthesis and
body composition, not cognitive outcomes. Also, without
an energy-balance control group for comparison, it is not
possible to separate the effects of the energy deficit from the
effects of learning or the environment.
When interpreting this study, it’s important to keep certain
caveats in mind. This study does not tell us how the dietary
protein-to-carbohydrate ratio would impact mood and
cognition in obese people or in individuals who diet while
already extremely lean. Nor does this study tell us how the
ratio would impact performance in physique athletes such as
bodybuilders or wrestlers. It also doesn’t tell us how the ratio
interacts with the absolute caloric deficit, as both groups in
the study had an identical 40% reduction in energy intake.
10
 nutritional counseling provided makes trials even harder to
The results of this study suggest that mood is compare.
affected during caloric restriction, regardless of
the specific macronutrient breakdown. However, Energy deficits have also been explored in several studies.

since the participants of this study were slight-
ly overweight young people, further research is
needed to confirm whether this association holds
true for other populations.
One found that two days of near 100% caloric deprivation
had no effect on mood in healthy young men, and a study
in physically active soldiers found a 30-day energy deficit
(40%) to not significantly affect mood-state in comparison
with a calorie-adequate diet.

In overweight and obese men and women, a six month long

The big picture
This study is one of many that examine differences in cog-
nition and mood between low- and high-carbohydrate
weight-loss diets that use different durations, study popula-
tions, and magnitude of energy deficit.
For instance, in an eight week study of overweight or obese
subjects, nearly all cognitive outcomes showed no difference
between low or high carbohydrate groups. In a separate
year-long study, obese subjects had similar mood improve-
ments from both low-carb and low-fat diets in the first two
months, but after a year
25% energy deficit was shown to improve depression scores
at three and six months, but participants who followed a
67% energy deficit for three months followed by a weight
maintenance diet only showed improvement at the end of
the intervention. In all three of these energy deficit studies,
results also showed no significant changes in cognition as a
result of dieting.
Perhaps surprisingly, energy deficits have often been shown
to not impact sleep. In the short-term near 100% caloric
deprivation study men-
tioned earlier, sleep was

  [...] overweight
the low-fat group had also unchanged, and a
greater improvements. previous study in over-
Cognitive outcomes
and obese people
weight premenopausal
weren’t different between women found that four
groups. weeks of an 800kcal/day

For these types of stud-

experience significant diet (energy deficit of
1460 kcal/day) did not
ies, outcomes have to
be taken with a grain
improvements on a affect sleep quality.

of salt. Mood improve-

ments may be due range of subjective Increasing the dietary
protein-to-carbohydrate

symptoms after weight

to weight loss rather ratio has been shown to
than diet composition. result in some cognitive
Study results also differ
loss, regardless of diet
improvements in other
depending on how mood studies, but typically for
was measured, and com- only certain selected
pliance with the diets
as well as the type of
composition. outcomes rather than
cognition or mood as

11
a whole. For example, one trial compared the effects of a
normal (1.5g/kg) or high (3g/kg) protein diet in healthy The results of this study support the idea that
normal-weight young adults. For three weeks, all foods and changes in mood may be influenced more by the
beverages were provided by the research staff. Cognitive duration and magnitude of a diet, rather than its
performance improved in both groups, but only the high macronutrient composition.
protein group significantly reduced reaction time. The
high-protein group ended up with significantly higher plas-
ma concentrations of the BCAAs, which the authors suggest
may have been the cause of the improved reaction time
Frequently Asked Questions
Would BCAA supplements impact these neurotransmitters?
Since BCAAs are the “problem” in regard to competition for
Collectively, the above suggests that changes in mood are
entry of neutral amino acids into the brain, it stands to rea-
reliant on a variety of factors, such as the length of the diet,
son that supplementing high doses of BCAAs would have
magnitude of caloric restriction, and occasionally com-
similar if not worse effects on brain chemistry.
position of the diet. Although the evidence is somewhat
mixed, these studies
A recent study evaluat-
also suggest that dieting
ed the administration

  Bottom line: Mood

may improve mood in
of a BCAA, essential
overweight/obese peo-
amino acids (EAA), or
ple more so than in lean
people. and depression issues a milk EAA mixture on
brain amino acid and

Importantly, while these

other studies had dif-
rarely have quick fixes neurotransmitter con-
centrations. It found that

in the form of pills.

the BCAAs in an amount
fering fat intakes, the
similar to many low
current study controlled
dose BCAA supplements
for fat intake so that the
that are currently being
only dietary manipula-
marketed does indeed
tion was the protein to carbohydrate ratio. It was also highly
blunt increases in serotonin by competitively inhibiting the
controlled, with subjects being housed in a metabolic ward
uptake of tryptophan from the bloodstream.
and given all meals and beverages. Thus we’re able to derive
a fairly specific conclusion: boosting protein probably won’t
Can I just take supplemental tryptophan to increase sero-
help you avoid the doldrums of starting a diet. The initial
tonin levels and improve my mood?
ten days of beginning a diet may be the most psychological-
If only it were so easy. While serotonin levels are associated
ly difficult, but that this will pass with time. Additionally, it
with mood, many other factors are involved. In studies where
appears that cognition and sleep is likely to remain relative-
people are purposely deprived of tryptophan, some show a
ly unchanged throughout a dieting period. It’s important
markedly worse mood. However, this is highly dependant on
to keep in mind that the results were in non-obese young
who the subject is, and those who have a history of depres-
people, so they can’t be generalized to athletes with much
sion are more likely to experience the effect.
greater caloric demands or to people who are pushing the
boundaries of leanness.
Tryptophan supplements have been widely studied, and do

12
improve mood (as well as sleep) in certain subjects. People
often face choices between medicating for moderate depres-
sion with SSRIs, or taking a supplement such as tyrosine or
5-HTP (the precursor to serotonin). Mood is an incredibly
complex phenomenon, and unfortunately studies rarely if ever
compare medications to supplements. And lifestyle factors
such as stress and sleep can be as or more important than what
you’re putting in your mouth.

Bottom line: Mood and depression issues rarely have quick

  Any mood
fixes in the form of pills. Supplements that can increase sero-
tonin can also have side effects from too much serotonin being
or cognitive
present throughout the body. It’s usually best to take a step
back and consider a variety of angles rather than a singular
disturbances
amino acid or other supplement.
that do occur
What should I know?
Jumping into a 40% energy deficit may result in mood distur-
appear to be
bances for the first ten or so days, which return to pre-dieting
levels by the third week, regardless of the protein-to-carbohy-
transient and
drate ratio. This matches up quite nicely with what many refer to
as the “low-carb flu”. The lack of differences between the groups may be owed
of the current study suggests that perhaps it should be called
the “dieting flu,” as it appears that these effects are not isolated to to the caloric
deficit itself
low-carbohydrate diets. Similarly, cognition and subjective sleep
quality showed minimal changes, some of which were transient.

When considered in conjunction with the benefits of higher

protein diets during times of caloric restriction for muscle
rather than the
preservation and satiety, it seems prudent to not worry about
the hypothetical problems that a high protein to carbohydrate
macronutrient
ratio would have on brain chemistry and neurotransmitters, as
these changes do not appear to significantly impact cognitive
ratios.
function. Any mood or cognitive disturbances that do occur
appear to be transient and may be owed to the caloric deficit
itself rather than the macronutrient ratios. ◆

In the mood for diet analysis? Join us at the Facebook ERD

forum to talk about low-carb flu, low-calorie blues, and how
the evidence matches experience.

13
 What If There Were No
Dietary Guidelines?
By Adele Hite
A recent episode of South Park lampooning America’s
epidemic of gluten anxiety features the Secretary of the
USDA agonizing over the realization that his agency has
been recommending Americans consume large quantities
of this “dangerous” substance as part of a high-carbohy-
drate, reduced-fat diet. Determined to make amends, an
animated version of Tom Vilsack asserts the importance
of the nutrition guidance his agency dispenses: “We are
the USDA! Without us, people would be eating dirt—and
chairs.” Vilsack ultimately saves the day by turning the
Food Pyramid upside down, and while some would applaud
this transformation—everyone ends up eating butter on a
stick—I have no fondness for the Food Pyramid in any ori-
entation, nor its replacement, MyPlate. Both shapes reflect
the advice of the Dietary Guidelines for Americans (DGA),
recommendations which divide the world of food up into
two groups: “healthy” food Americans should eat and
“unhealthy” food Americans should avoid. The basis for this
division has little to do with beneficial nutrients food may
or may not contain, but is instead based on the presence or
absence of components thought to impact chronic disease.
For the past 35 years, DGA guidance telling Americans
what to eat and not eat in order to prevent chronic disease
has remained remarkably consistent.
14
Problem is, it hasn’t worked very well.
During that time, rates of hypertension, high serum cho-
lesterol, and heart disease mortality have dropped. Some
researchers attribute these positive changes to dietary
“improvements” in line with DGA recommendations. Other
researchers blame the fact that obesity rates have doubled
and diabetes rates have tripled on our lack of adherence to
DGA recommendations. It's a neat trick, giving the DGA
credit for the good outcomes, but absolving them of the
negative ones.
If Americans have indeed
shifted their diets to align   It's a neat trick,
political document, and they regulate a vast array of federal
programs and services, influence health-related research,
and direct how food manufacturers respond to consumer
demand. Virtually no aspect of our food environment is
unaffected by the DGA. It is worth considering what impact
their absence would have on these other areas.
From the beginning, the DGA created clear “winners” and
“losers” in our food system. Winners were processed food
manufacturers who could reformulate products to meet
DGA standards; losers
were farmers who pro-
duced eggs and meat,
which couldn’t be easily

giving the DGA

more closely with the modified. When the DGA
DGA, the results are not directed consumers to
what was originally intend- avoid saturated fats, man-
ed. On the other hand,
if the U.S. government
credit for the good ufacturers replaced them
with trans fats. “Pink
has been unable, for the
past 35 years, to convince
outcomes, but slime” emerged from the
beef industry’s efforts
Americans to follow its
dietary advice, maybe it’s
absolving them of to produce lower fat
products.
time to quit trying.
the negative ones. Olestra, high-fructose
If there were no DGA tell- corn syrup, polydextrose,
ing Americans what to eat, soybean oil, and other
would heart disease mortality spike? Unlikely, since heart products of the food science laboratory were used to make
disease death rates had been in decline for over a decade food “healthier,” giving food manufacturers the opportunity
before the DGA were created. Would rates of obesity and to plaster health claims on labels and directing consumer
diabetes climb even further? Under current conditions, attention away from whole foods, which carry no labels and
i.e. DGA recommendations being followed or ignored no such claims. The disappearance of the DGA would help
depending on who you ask, the predictions are by 2030, level the playing field and perhaps begin to eliminate some
more than half of Americans will be obese and, by 2050, unwanted additives from our food supply.
one in three Americans will have diabetes. As obesity and
diabetes rates did not begin their rapid ascent until after Because the DGA influence research agendas, they have
the DGA were created, it is not likely eliminating them imposed their shape on emerging science; its outcomes
would make matters worse. reflect the policy in whose image it is made. Without the
DGA, government-funded nutrition science would operate
In fact, the DGA are far more than health prescriptions without the ideological constraints created by govern-
that Americans do or do not follow. They are a powerful ment-backed dietary advice. Without the DGA, a diabetes

15
prevention trial using a reduced-carbohydrate diet might not just
be an idle fantasy, but an NIH-funded reality.

Without the DGA, federal nutrition programs could tailor their

programs and practices to the needs of the individuals and com-
 Would
munities that they serve, instead of being directed by remote,
one-size-fits-all, top-down dietary guidance.
Americans
Finally, the DGA assert that science has unquestionably estab- starting eating
dirt—or chairs?
lished links between diet and chronic disease. It hasn’t, but the
DGA’s list of “good” and “bad” foods implies chronic disease
is entirely under the control of the consumer. This assumption
places the burden for prevention and cost of care on individuals,
relieving the government and other institutions of the responsi-
I don’t think so.
bility to improve economic, environmental, and social conditions
related to health. Eliminating the DGA would open up the possi-
Maybe, they
bility for rethinking this approach to public health.
might just start
In many ways, the DGA were a big fat mistake. It is time we
acknowledge that and move on, not by turning the current DGA eating—better.
“upside down,” but by throwing them out altogether. Would
Americans starting eating dirt—or chairs? I don’t think so. Maybe,
they might just start eating—better. ◆

Adele Hite is director and co-founder of Healthy Nation Coalition, a

non-profit health advocacy organization dedicated to promoting critical
nutrition literacy, individualized and community-based approaches to
food and essential nutrition, and an understanding of food-health rela-
tionships that goes beyond nutritional and caloric content of food. She
is also a registered dietitian and PhD student in communication, rheto-
ric, and digital media at North Carolina State University. She has masters
degrees in English education and public health nutrition and has pur-
sued graduate studies in nutrition epidemiology.

Prior to her graduate studies, she worked as the patient educator at Duke Lifestyle Medicine Clinic.
Her current research involves a critical examination of the U.S. Dietary Guidelines for Americans, par-
ticularly their history, their scientific underpinnings, their effects on the food-health environment,
and the implications of these recommendations as a structure of power in the sociopolitics of food
production and consumption.
16
The iPad Hangover
Evening use of light-emitting
eReaders negatively affects
sleep, circadian timing, and
next-morning alertness
Introduction
Within the last few decades there have been massive developments in portable technol-
ogy. High-powered devices have now become lightweight, convenient, and affordable.
Many activities, like book reading, have been digitized. With the development of this
technology, what once had been a common pastime to help get us to sleep may now
actually be doing the opposite by causing a shift in our circadian rhythm for sleep.

A circadian rhythm is essentially an organism's daily internal clock. In humans it

accounts for many of our physiological fluctuations throughout the day. A major mol-
ecule that affects our sleep biorhythms is melatonin. Many have heard of melatonin
used as a sleep aid, and for good reason. Melatonin is a hormone that is released by the
pineal gland in the brain and is involved with sleepiness and sleep regulation.

Melatonin production is heavily influenced by sunlight interacting with retinal pig-

ments. When light hits the retina, arylalkylamine N-acetyltransferase production is

17
depressed. Arylalkylamine N-acetyltransferase is an enzyme that catalyzes
a crucial step in melatonin biosynthesis. Therefore, when light is absent,
melatonin concentration builds and sleepiness ensues.

What happens when the retina is exposed to light during sleepy time hours?
Research has shown that exposure to artificial light at night suppresses mel-  The
International
atonin levels and increases alertness. When melatonin is suppressed, the
body is tricked into thinking it is still daytime and the circadian rhythm can
shift, especially when this happens repeatedly. This shift makes it difficult
to fall asleep. When a person can’t sleep due to (light-induced) melatonin
suppression, that signals there has been a shift in the biological clock, rel-
Agency for
ative to the normal 24-hour circadian cycle. As seen in Figure 1, levels of
hormones in the body such as melatonin and cortisol fluctuate throughout
Research on
the day. If the time course of one hormone is thrown out of whack, sleep
and other physiological outcomes may be shifted as well.
Cancer, an
Figure 1: Variation in melatonin and
agency
cortisol throughout the day
directly
related the
World Health
Organization,
has classified
shift-working
as a probable
carcinogen.
But why is this a problem? There is no definitive explanation as to why we
need sleep, but we know that chronic deprivation is detrimental to our
immune system, ability to perform, memory, and a laundry list of other
things in regard to general health.

18
Without question, sleep is important. In fact, chronic sup-
pression of melatonin via evening light exposure has serious
implicated health risks, such as cancer. Such conditions are
frequently seen in shift-workers, like nurses and fire fighters.
The International Agency for Research on Cancer, an agency
directly related the World Health Organization, has classified
shift-working as a probable carcinogen. Shift work represents
an extreme of altered sleep patterns, but the mechanisms
behind cancer incidence due to altered melatonin secretion
and circadian rhythm shifts have been widely researched.
Cancer risk aside, there are other legitimate reasons to study
this phenomena, as it directly relates to performance (both
mental and physical) and general health. The aim of this
study is to quantify the effects of light-induced melatonin
suppression, caused by iPad use before sleep, on sleep quali-
ty and feelings of lingering sleepiness after waking up.
Who and what was studied?
The title of this paper includes “light-emitting eReaders”,
but it is important to note that the experimental arm of this
study was conducted exclusively with Apple iPads.
This was an inpatient randomized crossover study that
included twelve healthy individuals who had all abstained
from stimulants (which was lab-verified), and were oth-
erwise free of any disorders that affected sleep quality and
sleep patterns. Three weeks before the study began, the
participants adhered to a strict sleep schedule of 10:00 p.m
to 6:00 a.m, which was verified by logs, call-ins, and wrist
actigraphy (which measures movement patterns).
The study participants lived on the premises for the duration of
the study. Sleep schedules were fixed at 10:00 p.m to 6:00 a.m.
For five days in a row, four hours before lights out, but in
otherwise dim light, the study participants either:
1. Used an Apple iPad on maximum brightness
2. Read a print book
Lighting conditions were kept tightly controlled through-
out the duration of the study. The iPad was at a fixed angle
and distance (about 1.5 feet away) from the participants
face to ensure equal light exposure between participants.
Study participants were able to pick whatever literature they
wanted, with the exception of comic books (possibly due to
different light reflections with colorful comics) and techni-
cal reading (possibly sleep inducing). Participants remained
in their beds at a fixed angle for the entirety of their reading
period, except for a 15 minute break at the 2.75 hour mark.
The research group measured:
1. How long it took the participants to fall asleep and
sleep activity.
a. Polysomnography (PSG) was performed on
  nights four and five. PSG records all biophysical
  activity and patterns during sleep.
b. Electroencephalography (EEG) concurrently
  measured brain activity.
2. Melatonin levels
a. On the fifth night, blood was drawn hourly and
  suppression was measured by comparison to
  starting measurements.
3. Alertness and sleepiness
a. Self reported: participants completed a dimly-lit
  computer generated survey both one hour before
  sleep and approximately six times during their first
  waking hour.
b. Physical measurements: EEG was taken upon
  waking under reproducible conditions. Study
  participants were asked to be still and stare at
  a black dot for 3 minutes. Riveting.
4. The group measured the spectral output of several
commercial e-reader devices at the same distance the
face was placed from the e-reader during the time of
the experiment.
19
 in a variety of conditions such as multiple sclerosis and
Study participants adhered to a strict waking and Alzheimer’s disease.
sleeping schedule during the study. Before bed,
each participant read for several hours, using Keeping in mind the suppressed melatonin levels and decreas-

either an iPad or a print book. es in REM sleep, iPad users reported lower levels of sleepiness
in the evening. EEG readings confirmed this by showing less
power in the delta/theta frequency ranges (associated with

What were the findings? sleep and drowsiness) when compared with print book read-
ers. The following morning, iPad users were much sleepier
iPad use consistently suppressed night-time levels of mel-
than print readers. It took iPad users on the order of hours to
atonin, which were significantly lower than when paper
“fully recover” from their sleep episode and gain full alertness
books were used. In fact, reading paper books resulted in no
after waking in the morning. Chronic use of these devices
suppression. Once the lights were dimmed, the melatonin
before bed may impair wakefulness, to the point where iPad
onset peak was shifted to a later time period with iPad use
users might not feel truly awake until late morning.
when compared to paper book use. Effectively, the iPad
artificially inflated the length of the “day” perceived by the
Data for the print book readers was essentially the oppo-
brain, by about an hour and a half! This shift in circadian
site, with greater feelings of tiredness in the evening and
rhythm for iPad users happened even though the room
increased wakefulness in the morning. This increase of
lights were dimmed.
tiredness in the morning experienced by iPad users is likely
related to decrease in REM sleep, and might be explained
These changes in melatonin were associated with changes in
by the circadian phase shift. Since the wake times were set
sleep patterns of the participants. On average, the iPad users
between the groups, and the phase was shifted to a later
took about 10 minutes longer to fall asleep when compared
time, the study participants who used iPads essentially woke
to paper book readers (at around 25 minutes, compared to
up when their body thought they should still be sleeping. If
15 in the paper book group).

Coincidentally, there was also a decrease of Rapid Eye

Movement (REM), or dream-inducing sleep, by about the   Effectively, the
same amount of time. However, the total amount of time
spent sleeping, the efficiency of that sleep (defined as the iPad artificially
inflated the length
percentage of time in bed spent actually asleep), and time
spent in non-REM sleep were not significantly different
between groups. While the exact function of REM sleep is
not fully known, one theory is that it’s involved in memo-
ry consolidation. As an analogy, computer hard drives can
of the “day”
accumulate fragmented segments of memory after long
periods of usage, and “defragmenting” the hard drive is a
perceived by the
function of the Windows operating system to remedy that
(or at least it was in the 2000s). A reduction in REM could
brain, by about an
have a multitude of impacts, including potential disruption
of the blood brain barrier, which is theorized to be involved hour and a half!
20
sleep and wake times were not regulated, it’s possible that this may not have been
observed. However, since most people must wake at a particular time (but not go
to bed at a particular time) it’s easy to see how these issues quickly compound.

The iPad’s peak light output is in the blue range of the visible spectrum at 452 nm.
The physiological effects from this blue light makes sense given the light output of
human’s original light source: sunlight also peaks under 500 nm, and melanopsin
pigment in the human eye absorbs in a peak range of about 480 nm. It’s not unre-
alistic to posit an “artificial sun” response when exposed to these higher energy
photons. Even though this wavelength may not seem as bright to humans when
exposed to an equal power of white light (which peaks at 612 nm), what mostly
matters to the brain is the excitation of the correct eye pigments to depress the
production of the arylalkylamine N-acetyltransferase, which in turn decreases
circulating melatonin.

In the short term, a person may get acute sleep onset insomnia due to a shift in
circadian rhythm. However, using electronic devices that emit blue light may
have biological ramifications in the long term, especially since chronic melatonin
suppression had been linked to an increased risk of various types of cancers.
Unlike in this study, though, most people self-select their bedtimes (and to a
lesser extent wake times). If a self-selected bedtime and wake time are not aligned
with the body’s natural circadian rhythm, this, combined with blue-light expo-
sure, may further exacerbate the phase-shift phenomenon and lead to chronic
issues associated with sleep deprivation.

Lastly, a four-hour read time was established in this study, when many people
may spend even longer times in front of a screen (either TV, computer, tablet,
or phone) from afternoon to bedtime. The effects of longer exposure times are
unknown, while tablet and phone displays continue to get brighter each year.

Study participants using iPads before bed took about ten minutes
longer to fall asleep, experienced less REM sleep, and found it much
harder to feel fully awake in the morning than people that read print
books before bed.

The big picture

The findings are somewhat scary for the average person who loves the Internet,
which houses a vast collection of research on nutrition and health issues (and cat
pictures) that can easily keep you up for hours. Back-lit tablets suppress and delay

21
melatonin, affect EEG, negatively affect sleep quality, and
hinder the ability to wake up in the morning. But do the
findings of this rigorously controlled study apply to free-liv-
ing humans?
At first glance, it may seem difficult to translate the data
to the typical person not staying in such controlled con-
ditions. But this is not the case. On the one hand, study
participants remained in a bed (in a reclined position) for
almost the entire duration of the study with an IV insert-
ed and electrodes stuck to their heads. On the other hand,
this may actually be a decent model system for the typical
inactive adult.
  If you have to
use a tablet/
laptop/phone at
night, it may be
helpful to dim
the screen as low
as possible, while
still being able
to read.
Assuming a person works for eight hours per day at a desk
(staring at a light emitting screen), sleeps for eight hours per
day (but probably less) and then spends four plus hours per
day using electronic devices for entertainment rather than
work, it’s probably not much different than the lab setting.
The study participants also had to adhere to regular eating
and sleeping times. Taking these things into account, it is
likely that the negative effects of this study may even be
understated. This study only had people in a phase shift-
ed state for five days, where a person may be chronically
phase shifting for years. This may lead to chronic infections,
general malaise, premature fatigue/overtraining, and even
precursors to metabolic syndrome.
The use of blue-light emitting devices is growing at a consis-
tent rate. The use of these devices near bed time is likely to
interfere with the body’s natural sleep patterns. Continued
interference may lead to chronic sleep issues, which are
linked to a host of negative conditions listed above. The
study is not clear about how quickly this phase shifting
takes place and how quickly it can be corrected. Since there
are quantifiable differences between the two experimental
groups, and both groups went through each condition, it is
safe to say that these phase transitions can be altered over
the course of days.
Take note that there is nothing uniquely bad about iPads or
even tablets in general. Phones, laptops, and televisions are
all devices that emit blue light at substantial levels.
Frequently asked questions:
If I wanted to time my iPad use to avoid a phase shift, what
is recommended?
If a four hour exposure (from 6:00 p.m. to 10:00 p.m.) leads
to a 1.5 hour phase shift, then presumably a substantially
shorter exposure time ending much earlier in the evening
would lead to less phase shifting and more appropriately
timed melatonin secretion. But further research would be
needed to know specifics. Keep in mind that if the four hour
exposure occurred much earlier, in the daytime hours, mel-
atonin would already be suppressed by sun/light exposure
and you wouldn’t have to worry.
If you have to use a tablet/laptop/phone at night, it may
be helpful to dim the screen as low as possible, while still
being able to read. One way to get around this is using blue-
22
light blocking glasses, which have amber lenses and hence
make you look quite silly (but also smart and innovative
compared to friends/mates that may mock you.) As seen
in Figure 2 these have been shown to allow for melatonin
production, and hence improve sleep quality and potentially
even mood. If you can’t or won’t use these glasses, but still
want to use your device at night, consider free programs
such as “f.lux” for laptops or other alternatives for mobile
devices. These automatically make the screen dimmer and
more red as the sun sets. Some apps can even lower screen
brightness more than a device allows on its own.
Can supplementing melatonin help to restore altered sleep
rhythm?
If melatonin is used as a sleep aid in order to fall asleep
during the time of maximum sleep propensity, this is plausi-
ble. Examine.com has a very extensive article on melatonin
and its proper supplementation protocol. Melatonin may be
an option for artificially attenuating the down-shifted mel-
atonin peak if an iPad is used before bed. That being said,
Figure 2: Effect of light and blue-light blocking glasses on melatonin
melatonin isn’t the only determinant of circadian rhythm,
so artificially restoring melatonin levels probably won’t
eliminate all the negative effects of bright electronic devices.
So while the specific efficacy of melatonin for phase shift
caused by iPads hasn’t been researched, there is a cheaper
and simpler solution. Occam’s razor dictates that the sim-
plest answer is often correct -- rather than putting a bandaid
on the problem of shifted sleep, you could work on enforcing
a habit of natural light exposure. Some sunlight during the
day, and no bright light in the hour or more before bed. It
sounds tough, but starting slow (for example, shifting device
use a bit earlier each week) can make the transition easier.
The study authors noted that the relatively easier time print-
ed-book readers had in falling asleep is actually similar to
the effect size of the popular insomnia medication Lunesta.
Simply constraining bright light to earlier hours could be an
effective and side effect free treatment option for those with
sleep issues.
23
   Even being exposed to low intensity
blue light, like those blue LEDs on power
cords, can make you more drowsy the
next morning.
Should I have any issues with my e-ink device? Will using a blue-light emitting device during the day inter-
No, the “spectrum” of the e-ink device is comparable to a fere with my sleep patterns?
regular paperback book. However, some e-ink devices now This is unlikely, as the mechanism of this phenomena is
have front-lit capacity, where light is directed on to the associated with delayed melatonin accumulation. Melatonin
screen, rather than projected through it. There is no peer is at imperceptible levels during the day, so there would be
reviewed data on spectral emission from those devices in no quantifiable effect of inhibiting its synthesis with blue
this study, but anecdotally some of these devices appear to light. The sun emits the very same wavelengths in apprecia-
have somewhat bluish hues (as opposed to the yellow of cer- ble amount.
tain lightbulbs). Reading these devices on a dim brightness
setting with room lights turned off may be comparable or
even better than reading a print book with dim room light-
What should I know?
Most all modern back-lit portable electronic devices emit
ing, since less of your field of vision will be lit. This has not
high-energy blue light. The human eye is sensitive to blue
been formally studied though.
light exposure, leading to suppressed melatonin production.
This suppression of melatonin production causes a shift in
Bulbs that can change color and brightness automatically,
sleep patterns. Hence, sleeping out of line with your natural
such as the Philips Hue, are increasingly popular. While you
endogenous sleep patterns may lead to chronic sleep defi-
don’t have to go out and buy fancy smartphone-controlled
ciency. Chronic sleep deficiency and consistently impaired
light bulbs (although they are kind of cool), at least dim-
melatonin production may increase the risk of many incon-
ming your room lights at night can greatly help melatonin
veniences, disorders and diseases.
secretion. What about the lucky people who can sleep with
lights on, maybe falling asleep to the TV or while your part-
While other health topics (such as low carb vs high carb) may
ner is awake? They’re not so lucky after all -- sleeping with
be subject to many more studies, one of the most import-
the lights on causes sleep to be shallow with more likely
ant but undercovered health issues may be the downstream
arousals in the middle of the night.
effects of poor sleep caused by using these devices at night. ◆
Even being exposed to low intensity blue light, like those
blue LEDs on power cords, can make you more drowsy the Don’t you dare read this past 10pm. But if you choose to
next morning. So it can pay dividends to be more aware of anyway, head over to the ERD private Facebook group to
what kind of light surrounds you throughout the evening all talk about how to improve your light-related sleep habits.
the way into slumber time.

24
 Can mice get
cancer from
steak?
A red meat-derived
glycan promotes
inflammation and
cancer progression
Introduction proof to hold any of these reasons up as the definitive cause
of the increased cancer risk.
Red meat is one of the foods the media loves to hate, per-
haps while still secretly loving it. Any time new research
Although there is no definitive proof, the collection of
is published that deals with red meat consumption and X
mechanisms, human biomarker RCTs, animal studies, and
(where X might be cancer, heart disease, diabetes, etc.) the
observational studies do see to support the hypothesis that
major news outlets pick it up, running scary headlines to
red meat increases digestive cancer risk. Since there proba-
attract clicks. The science, however, is typically a bit more
bly will never be a human RCT with red meat as a variable
nuanced than the bold claims of news headlines would lead
and colorectal cancer as an outcome, we’re unlikely to be
the average reader to believe. So is there actually anything
able to assess causality in humans.
to worry about when eating red meat?

This particular paper attempted to explain one possible

There is a modest but consistent correlation between diets
mechanism involved in the meat-cancer correlation from
high in red meat and cancer risk in human epidemiological
an immune response standpoint. Our simplified images
studies. Scientists haven’t been able to figure out why this
of a cell typically depict a roundish circle denoting the cell
link exists yet, or if the risk is increased by red meat directly
membrane, with a bunch of stuff inside the cell. In reality,
or through some other related factor. A number of hypothe-
cell surfaces are covered with different kinds of mole-
ses have been proposed: grilling red meat creates dangerous
cules sticking out from the cell membrane. Many of these
compounds, the iron in red meat generates free radicals, or
molecules are long sugar chains made up of units called
people that eat more red meat are more likely to engage in
monosaccharides. The researchers looked at a particular
other diet and health behaviors that promote cancer, such
group of monosaccharides called sialic acids, which are
as a high fat or low vegetable diet (which is largely based on
linked into larger structures known as glycans.
correlational findings). So far, there isn’t enough substantive

25
The two sialic acids of interest were n-acetylneuraminic acid
(denoted Neu5Ac) and n-glycolylneuraminic acid (denoted
Neu5Gc), which are seen in Figure 1. Structurally, these two
molecules are identical except for an additional hydroxyl
(-OH) group on Neu5Gc. These molecules are found in a lot
of animal-based dietary sources in different levels, with beef
and caviar having the highest levels of Neu5Gc.
Neu5Gc is interesting, because it’s not something that
humans can synthesize from Neu5Ac - the gene that codes
for the enzyme needed to add on the hydroxyl group had
a sequence deletion several million years ago and thus is
nonfunctional. This likely conferred several evolutionary
advantages. First, it may have been a factor in our ability to
increase our brain development beyond that available to our
primate cousins. Second, certain pathogens such as select
strains of malaria as well as an E. coli toxin bind to Neu5Gc,
so lacking the enzyme necessary to synthesize Neu5Gc
Figure 1: Humans and Neu5Gc
would have provided a resistance advantage for these dis-
eases many millennia ago.
Despite this genetic change, Neu5Gc is still found in low
levels in normal human tissue - so any molecules would
have to come from dietary sources and then be incorporat-
ed into our own cell membranes. It’s also found in higher
levels in malignant tissue. The researchers speculated
that because this molecule is “foreign” to our bodies, our
immune systems would react to it by producing antibodies
to attack it. Indeed, it has been shown that 85% of people
tested had antibodies against Neu5Gc. It is not known why
15% of the people tested did not possess the antibodies.
An immune system that is constantly turned “on” results in
inflammation. Markers of inflammation increase when the
immune system is fighting off a foreign pathogen. In acute
situations, inflammation is beneficial! However, we’re learn-
26
ing that chronic inflammation does not do a body good. Figure 2: Neu5Gc in common foods
Chronic inflammation has been linked to heart disease,
various bowel syndromes such as irritable bowel syndrome
(IBS) and Crohn’s, and yes, cancer.

So, we finally have a hypothesis: A) red meat consumption

leads to B) incorporation of a foreign molecule into our
bodies, which leads to C) an immune response to the for-
eign molecule, which leads to D) systemic inflammation,
which leads to E) an increased incidence of cancer. The
researchers used simulations of A and C, and evaluated
their hypothesis in mice.

Though the media often links red meat consump-

tion to an increased risk of cancer, scientists are still
trying to determine what aspect of red meat results
in cancer, or if this link exists at all. Researchers in
this study tried to determine if red meat consump-
tion results in systemic inflammation due to the
presence of foreign, meat-derived molecules.
phospho-N-acetylneuraminic acid hydroxylase (CMAH)
enzyme needed to convert Neu5Ac to Neu5Gc. Unlike most
Who and what was studied? of our mammalian friends, including primates that are still
To confirm and quantify previous findings, the research- very genetically similar to us in other ways, humans do not
ers used a new analytic method to determine the amount produce a functional CMAH enzyme due to a deletion in
of Neu5Gc in different foods. The results are summarized the gene sequence for the enzyme. So like in humans, any
in Figure 2. Neu5Gc was absent in fruits and vegetables, Neu5Gc in these genetically modified mice would have to
most seafood, poultry, and butter. It was found in low lev- come from dietary sources.
els in milk and cheeses, bison, lamb, and pork. In beef, the
Neu5Gc content was approximately 2-30 times higher than
Keeping the notations straight: All gene symbols
in other red meats and ranged from 0.025-.231mg/g, the
are italicized (ex. Cmah) while all protein abbrevi-
highest of any of the tested foods other than caviar. The
ations are all caps and non-italicized (ex. CMAH).
Neu5Gc content of beef varied greatly however, and the
lower range was more in line with other foods. By contrast, Human gene symbols are written in all caps, while
poultry, dairy, and seafood were all high in Neu5Ac. mouse and rat gene symbols capitalize only the
first letter. So the full length mouse Cmah DNA
The researchers used a genetically modified mouse mod- sequence produces the CMAH enzyme, while the
el for their experiments, as is often done in initial in vivo truncated human CMAH DNA sequence would
studies. This particular mouse had the Cmah gene removed, produce no functional enzyme.
meaning that they could not produce the cytidine mono-

27
   [...] most epidemiological data in
humans has linked diets high in red
meat to an increased risk of colon and
colorectal cancer. [...] the organ of interest
for spontaneous tumor formation was
the liver.
Male mice were fed a normal diet, or a normal diet supple- animals to mount their own immune response against the
mented with porcine submaxillary mucin (PSM). Translation: Neu5Gc.
pig salivary gland excretions, yum. PSM has a high
concentration of Neu5Gc and was meant to simulate the con- Two additional experiments created the anti-Neu5Gc
sumption of red meat in humans. The PSM-supplemented immune response prior to feeding and looked at HCC
mouse diet contained 0.25mg of Neu5Gc per gram of feed, tumor formation. One tested the specificity of the immune
which is similar to the highest concentrations seen in red response by feeding the mice the PMC enhanced diet,
meat. After being fed this diet for several weeks, the levels which created elevated levels of Neu5Gc, or a diet enhanced
of Neu5Gc in the mouse were similar to levels observed in with the unhydroxylated precursor molecule Neu5Ac. A
humans adhering to a red meat rich diet for several years. second experiment compared pre-immunized wild-type
mice that had a functional Cmah gene to the genetically
It’s important to note that most epidemiological data in modified mice without the Cmah gene.
humans has linked diets high in red meat to an increased
risk of colon and colorectal cancer. In this particular mouse
model, however, the organ of interest for spontaneous tumor
What were the findings?
The first experiment showed that elevated markers of
formation was the liver (hepatocellular carcinoma, or HCC).
inflammation were only seen in the mice that were fed
The researchers used liver cancer as an outcome because this
PSM-enhanced food for 12 weeks prior to receiving an
mouse strain is naturally susceptible to spontaneous liver
injection of Neu5Gc antigen. Compared to mice that were
tumors. So the choice of liver cancer was somewhat arbitrary
fed standard food for the same time period, or fed PSM-
-- if the strain had been susceptible to pancreatic cancer, they
enhanced food but received a control injection, levels of
may have used that as the outcome instead.
interleukin-6, serum amyloid A protein, and haptoglobin
were all elevated at highly statistically significant levels. This
To create models of inflammation similar to humans, PSM-
was also confirmed to happen in a dose-dependent manner;
and control-fed mice were immunized against Neu5Gc by
as the injected antigen concentration increased, so did the
injecting red blood cell ghosts (which are empty red blood
levels of inflammation markers.
cells) that did or did not contain Neu5Gc. This caused the

28

  Scientists crack
why red meat is
linked with cancer
- and SUGAR may
be to blame.
- Headline on the Daily Mail, 12/30/14
In the second experiment, the test mice were first exposed
to Neu5Gc immunogens (which are able to induce an
immune response in addition to an antibody response)
through injections of red blood cell (RBC) ghosts. The mice
were then fed either PSM-enhanced diets, or diets rich in
Neu5Ac through supplementation with edible bird’s nest
(EBN). Translation: the salivary secretions of the swiftlet
bird, highly prized as a culinary delicacy. Yum?
Because the mice lacked the Cmah gene, they were unable
to convert the dietary Neu5Ac to Neu5Gc. Here, elevated
interleukin-6 levels were seen only in the PSM-fed mice
and not in the EBN-fed mice. Additionally, 2 out of 7 mice
(29%) in the PSM feed group developed HCCs over the
course of 55 weeks, whereas none of the mice in the EBN
feed group developed tumors.
The final experiment compared the genetically modified
mice with their wild counterparts. Mice were pre-ex-
posed to either chimpanzee RBC ghosts with cell surface
Neu5Gc immunogens, or human RBC ghosts with cell
surface Neu5Ac immunogens. All mice then received the
PSM-enhanced food. After 85 weeks, a comparable num-
ber of wild type and genetically modified mice that had
been exposed to the Neu5Ac immunogen had developed
HCCs (9% and 7% respectively) but 47% of the genetically
modified mice exposed to the Neu5Gc immunogen had
developed HCCs.
Mice fed diets high in a cell surface sugar mol-
ecule found in red meat had higher levels of
inflammation markers. The inflammatory
response was specific to Neu5Gc and not the simi-
lar Neu5Ac precursor molecule. The combination
of consuming that sugar molecule and manipulat-
ing the mouse immune system to make antibodies
against that sugar molecule caused mice to devel-
op more liver tumors than control groups.
The big picture
“Scientists crack why red meat is linked with cancer - and
SUGAR may be to blame” - Headline on the Daily Mail,
12/30/14
Contrary to media assertions claiming that scientists now
know why red meat causes cancer, this study actually had
some limitations. First off - there is no human data present-
ed. There is an interesting mechanism proposed and a very
carefully designed model created to ask a specific mecha-
nistic question, but this may just be the first baby step on
a long journey to “knowing” the true cause, if a single one
even exists, for the increased incidence of cancer seen in
consumers of red meat in large quantities. Mice models are
almost always the first starting point for asking a question
about biological cause and effect, but results should nev-
er be taken as conclusive or directly applicable to humans
at this stage. Cancer has been caused and cured in mouse
models a dozen times over, in ways that have turned out to
be completely inapplicable in humans.
The study design itself, aside from being only conducted in
mice, had its own limitations. Studies in genetically similar
mice provide nowhere near the applicability of data that a
29
large scale human study that includes a wide variety of population factors (age,
sex, race, weight) would. Also, the PSM-enhanced food intervention was meant
to simulate only red meat consumption, not to simulate a typical human diet that
is usually full of varied food products other than red meat. For example, a recent
small study in humans suggested that consumption of a certain type of starch
may counteract some of the cancer risk effects of red meat consumption. And the
immune exposure to the Neu5Gc immunogen in the final experiment introduced
an external factor at a single point in time, instead of a lifetime of immune system
exposure that likely has varied and currently unknown effects. The immune sys-
tem is incredibly complex, and there are examples of immune system responses
having both pro- and anti-cancer effects.

As the study authors note, there is currently no published research that examines
the correlation of anti-Neu5Gc antibodies in humans with cancer risk. So it will
be important for future research to examine the mechanism found in this study
in a human sample. Finally, there is the question of organ system relevance. This
particular mouse model was used because there is a low level background inci-
dence of spontaneous tumor formation in the liver over time, which more closely
resembles the spontaneous formation of tumors in older humans rather than
tumors that are artificially implanted in mouse organs for research purposes.

However, red meat consumption has not been correlated with increased liver
cancers. Some of the primary non-genetic risk factors for that disease are certain
viral infections and alcohol consumption, both of which may also involve some
aspect of chronic inflammation, but have never been linked to Neu5Gc. Cancer is
a complex disorder, so it’s difficult to draw direct comparisons between a tumor
that forms in the liver in one species versus one that forms in the colon in anoth-
er species. The tumors in the mouse did show incorporation of Neu5Gc into the
tissue, but Neu5Gc has been found in a number of different malignant tissues in
humans (as well as in healthy tissue), so there may be non-specific incorporation
of the molecule into tumors in general.

The results of this study suggest a possibly novel explanation for the correlation
between red meat consumption and cancer, and might be able to explain some of
the contradictions with previously proposed explanations. The jury is still out on
whether this is relevant to humans. The research team believes in the importance
of this study, however. The last two authors listed on the paper have licensed
anti-Neu5Gc antibody technology from their institution and co-founded a com-
pany that is investigating targeted antibodies as a potential cancer therapeutic. It’s
possible that regardless of any mechanistic explanations, Neu5Gc still might be
an interesting new target for cancer drugs, or perhaps a biomarker that can be
used in additional research.
30

This study provides evidence to support a molec-
ular link between meat consumption and cancer
caused by inflammation, but since no human
evidence was gathered in this trial, much more
research is needed before these results can be
applied to people.
Frequently asked questions
How applicable are these results to humans?
The fact that the genetically modified mice lacked the gene
that is also non-functioning in humans did make this
mouse model more “human-like” for the purposes of this
study’s aims. The quantity of the food exposure is at least
  Cancer has been caused and cured in
mouse models a dozen times over, in
ways that have turned out to be
completely inapplicable in humans.
somewhat similar to human intakes, unlike some studies
that show that a particular chemical or food is carcino-
genic when fed to rats in 100 or 1000 times the quantity
a human could ever possibly consume. That being said,
it would be nearly impossible for a human to consume
a 100% beef diet consisting of the highest-Neu5Gc beef,
except for the small number of people who eat only red
meat and no plant products.
There are several reasons why the results may not apply to
humans. Mice were consuming only the enriched feed in
the experiments, whereas humans typically have a much
more varied diet that does not include only red meat. Also,
no other related factors such as other dietary patterns and
exercise were evaluated. And then there’s the question of the
differing effects of a sudden immune trigger by introduc-
ing exposure to an antigen or immunogen, compared to a
lifelong antibody exposure in humans that likely begins as
soon as a child begins eating solid foods. Finally, the organ
of interest in the mice was completely different than the
organ of interest in humans, so drawing any direct compari-
sons at this point is challenging.
Should I change my diet?
Probably not solely as a result of this study. While there
are some interesting mechanistic correlations in the data,
there is too little evidence to be able to draw conclusions
from these experiments alone. This particular study was
not attempting to add to the body of knowledge that shows
a slight but consistent correlation between red meat con-
sumption and elevated cancer risk in humans, but rather
attempting to tease out a possible molecular cause for the
effects seen, given that previous theories such as mutagens
created by grilling and generation of free radicals may not
explain the whole story.
Even one of the study authors commented in an interview
that dietary changes are probably not necessary for every-
body, and if this line of research does bear out it may be
more useful for people with either a previous personal his-
tory of cancer, or a family history of cancer.
31
Should I be concerned about inflammation? What can I do
about it?
What should I know?
In a very carefully designed and controlled experiment,
Possibly - a lot of studies have shown that chronic inflam-
genetically modified mice were fed a diet meant to simulate
mation in general has a lot of negative health effects. As
consumption of red meat (and only consumption of red
mentioned previously, even low levels of inflammation have
meat, rather than red meat in the context of a much broad-
previously been correlated with a number of negative health
er and varied diet). After being exposed to an immune
effects. Regardless of any specific connections that may come
challenge against a cell surface molecule found in red meat,
from further inquiry of this data, making efforts to reduce
the mice had increased markers of inflammation and an
systemic inflammation is likely a healthy long term choice.
increased occurrence of liver tumors.

Quite a few supplements have been touted as reducing

More research is warranted into this possible mechanism, as
inflammation, but here too much more research is needed.
there isn’t yet human data, so applying these results directly
A number of diet and lifestyle changes can be easily made
to humans is more speculative at this time. In terms of red
though. Many of the things that reduce inflammation, such
meat and protein in general, Examine.com has looked at the
as reducing tobacco and excessive alcohol use, improv-
human science behind the observed increase in cancer risk
ing sleep, consuming more plant-based foods, increasing
exercise and utilizing stress reduction techniques, promote
multiple times, most recently here and here. ◆
general health outside of any inflammation concerns as well.
To discuss mouse models, in vivo experiments, and recipes
with and without red meat, visit the ERD Private Forum on
Facebook.

32
 Sodium phosphate:
a potentially underutilized
ergogenic aid
Effect of sodium phosphate supplementation
on repeated high-intensity cycling efforts
Introduction Sodium phosphate has shown promise as an ergogenic aid
for endurance athletes thanks to a growing body of research,
Phosphate doesn’t get much attention in the nutrition and
though it is relatively unknown whether supplementation
supplement world, but it performs many important func-
improves performance outside of a lab, in the context of the
tions during exercise and everyday life. These include acting
ever-changing demands of athletic competitions. Previous
as an intracellular buffer and enabling the release of oxygen
research has found improvements in maximal aerobic
from hemoglobin. It also plays a critical role in energy pro-
capacity (VO2 max) after supplementation, though mixed
duction as a part of the basic structure of phosphocreatine
results have been found for lower intensity endurance
and adenosine triphosphate (ATP). Phosphocreatine serves
performance.
as an energy store that is available immediately during exer-
cise, and can release energy without the need for oxygen.
The purpose of the study under review was to investigate
ATP contains three phosphate groups and is often referred
the effects of sodium phosphate supplementation in a
to as the body’s energy currency.

33
research setting that's a more realistic reflection of real-life cycling condi-
tions (repeated maximal sprint and short time-trial efforts), after six days Why do study
of sodium phosphate supplementation, as well as four days after stopping
results differ for a
supplementation, to determine any lasting effects.
given ergogenic
Who and what was studied? supplement?
Seventeen competitive male cyclists with an average age of 34.7 completed
One reason that sports supplements
this study. As seen in Figure 1, these are elite level athletes who ride about
can show benefit in some studies
200 miles per week, with an average of 5.4 years of racing experience and
but not others is because there are
VO2 max of 71 ml/kg/min.
many different testing procedures
that researchers can use to deter-
Figure 1: Elite male competitive cyclists in the study mine if there is an effect from the
supplement trial.

For example, studies can test

participants using a time-trial (cov-
ering a pre-set distance as quickly
as possible), time to exhaustion
(maintaining a pre-set pace for as
long as possible), or repeated sprint
tests (generating as much power as
possible each time). Within these
different protocols there is a lot of
potential variability, like the dis-
tance of the time trials, the intensity
for the time to exhaustion tests, or
After an initial familiarization session and assessment, baseline perfor- the number of repeated sprints to
mance was measured, followed by six days of supplementation with either be measured, as well as rest time
sodium phosphate or a placebo. The performance testing was then repeated between efforts, whether the partic-
on day one and day four after stopping the supplementation. ipants are recreational athletes or
trained professionals, etc.
This was a double-blind placebo-controlled study. Participants were giv-
en either 50 milligrams per kilogram of fat-free mass per day (about 3.3 Underlying all of these variables
grams) of sodium phosphate or placebo (mix of glucose and table salt), are gender differences, as men and
divided into four doses and taken with meals (every four to five hours) for women can differ significantly when
six days. it comes to fuel sources, metabo-
lism, and specific adaptations to
Great care was taken to replicate the potential confounding variables before exercise. And most studies, including
each test. Participants did not exercise for 24 hours before each trial, and this one, have been in males only.
testing took place at the same time of day (within an hour) to control for

34
circadian variation. Participants also kept a 24-hour diary
of all food and drink intake prior to the initial test and were
required to replicate that energy intake as closely as possible.
The test protocol was modified from cycling race simula-
tions used in previous research, and attempted to emulate
the constant up and down effort of actual racing. This test
took 43 minutes to complete, which is substantially shorter
than cycling road races. The test was, however, similar to
criterium and certain track cycling races.
The protocol itself was a mix of four sets of 6 × 15 second
maximal sprints (with 45 or 15 seconds recovery in between
sprints) and two sets of five minute time-trials. Active
recovery for three minutes (pedaling at 100 W) separated
each set.
Seventeen competitive male cyclists supplement-
ed sodium phosphate or a placebo for six days.
Performance tests were done on the first and
fourth day after supplementation stopped.
as overall power output during sprints (5%) and time trials
(4%) in the sodium phosphate group. Supplementation had
no impact on performance during the first set, but it led to
performance improvements in the second, fourth and fifth
set, effectively reducing the fatigue seen in the later sets
of sprints. No differences were seen in the placebo group
during any set of sprints.
Similarly, supplementation led to an improvement of 9%
compared to baseline during the second time trial, which
was set six. It is hard to say why the power output in set six
was higher than set three, but it may be due to psychologi-
cal factors related to the final section of the test. Again, no
differences were seen in the placebo group during either of
the time trial sets.
Another important aspect of this study is that no differences
in power output were found between day one and day four
in the sodium phosphate group, meaning the effects of the
supplementation were still present four days after supple-
mentation had ceased. No changes in power output were
observed in the placebo group throughout the testing.

No differences in serum phosphate concentrations were

What were the findings? reported between the groups at baseline, day one, or day
Increases were seen for overall power output (5%), as well four. However, compared with baseline, serum phosphate

What the heck is the “smallest worthwhile change”?

P-values are used in research in order to establish statistical significance, i.e. the
probability of whether or not a relationship between two variables has occurred
by chance and how strong the relationship is. But often there isn’t a way to make
inferences about the clinical or real-world significance of the effect.

Sometimes, useful effects may not be statistically significant, while statistically

significant effects may not always be actually useful. Using an alternative sta-
tistical model, researchers can establish the smallest change that would still be
beneficial (or harmful) and use the confidence limits to make a qualitative state-
ment about the real-world significance of the changes. This study also classified
outcomes for the smallest worthwhile change as beneficial, trivial, or harmful.

35
   [...] it is reasonable to think that
sodium phosphate would benefit
endurance athletes competing in events
lasting not only from 15 minutes to one
hour, but likely into the four to seven hour
window of elite road races.
actually decreased in the sodium phosphate group on day
one and day four, while serum phosphate concentrations in
the placebo group remained unchanged.
Total work (measured in kilojoules) increased in the sodium
phosphate group on day four, while the increases seen on
day one were not statistically significant. However, analysis
for the smallest worthwhile change indicated that improve-
ments in overall power output and overall sprint power
output (but not time trial power output) would likely still
be of benefit in competition. No changes were seen in the
placebo group.
Cyclists supplementing sodium phosphate expe-
rienced a 5% boost to power output during the
performance test. Most of the benefits were
observed later in the tests, suggesting sodium
phosphate can reduce fatigue.
What does the study really
tell us?
This study adds to the growing evidence that suggests
sodium phosphate may be an effective ergogenic aid for
endurance athletes. Six days of supplementation resulted in
improved performance (greater work and power outputs)
during a simulated cycling road race, which included a mix
of repeated short-duration sprints and five-minute maximal
efforts. Impressively, the benefits still existed four days after
supplementation had stopped.
Since the test used in this study took 43 minutes to com-
plete, it is unclear how these benefits would translate to
a road cycling race lasting three to seven hours. However,
because of the lingering effects of the supplementation, it
is reasonable to think that sodium phosphate would bene-
fit endurance athletes competing in events lasting not only
from 15 minutes to one hour, but likely into the four to
seven hour window of elite road races.
There are a number of proposed mechanisms behind these
improvements, which the study did not specifically explore.
These include:
• Increased phosphate availability contributing to cre-
atine phosphate synthesis, phosphate-stimulated
glycolysis, and enhanced oxidative metabolism (even
with no changes in serum phosphate levels).
• Enhanced red blood cell 2,3-diphosphoglycerate
36
 (2,3-DPG) synthesis, promoting a decreased affinity
of hemoglobin for oxygen and resulting in great-
er unloading of oxygen to the peripheral tissues.
Previous research has shown that it took approxi-
mately two weeks for RBC 2,3-DPG concentration to
return to baseline following only three days of sodium
phosphate loading.
• Improved myocardial efficiency, providing a great-
er stroke volume (and cardiac output), leading to
increased and more efficient oxygenation of the mus-
cles during exercise.
• Improved hydrogen ion buffering capacity, which is
important for repeated sprint ability.
All of these factors may have played a role in this study, con-
sidering this protocol would have engaged multiple energy
systems (creatine phosphate, anaerobic glycolysis, aerobic
glycolysis) during the course of the test.
This study adds to the growing body of evidence
that supports the use of sodium phosphate as an
ergogenic aid for endurance athletes.
The big picture
“Six days of sodium phosphate supplementation resulted in
enhanced performance during a simulated high-intensity
  The differences here were small (about
5% improvements) but highly significant
in the world of competitive sports, where
fractions of a second can make the
difference between winning and losing.
road cycling protocol incorporating repeated-sprints and
short duration time-trial efforts. These benefits were still
evident four days after supplementation had finished, with
no performance differences found between day four and
day one post-loading.”
The differences here were small (about 5% improvements)
but highly significant in the world of competitive sports,
where fractions of a second can make the difference
between winning and losing. Sodium phosphate appears to
be an effective supplement for trained endurance cycling
athletes looking for improvements.
There are still many unknowns in regard to sodium phos-
phate supplementation, such as how long the optimal
dosing period is, how long the effects will last upon stop-
ping the supplement, effects of repeated loading periods,
whether gender differences exist, and precise mechanisms
of action.
Keep in mind that energy demands differ between sports,
and sodium phosphate supplementation may affect athletes
in various non-cycling endurance sports, such as running
and swimming, differently.
Frequently Asked Questions
What type of sodium phosphate did the study use?
This study (and many others) used tribasic dodecahydrate
37
sodium phosphate. Previous research using dicalcium phos-
phate or calcium phosphate has failed to find performance
improvements. It is possible that the lack of performance
enhancements were due to using a different loading pro-
tocol (in some cases only giving a single dose), or possibly
lower bioavailability. However, further research is needed to
confirm this hypothesis.

How much sodium phosphate should I supplement?

Studies have generally used three to five grams or 50 mil-
ligrams per kilogram of fat-free mass, per day. However,
different dosages haven’t been compared much against each
other. And as mentioned previously, studies tend to be in
males only. Along with the limited sample size of many stud-
ies, and their inclusion of highly-trained endurance athletes,
the results may not apply as well to a typical active individual.

Will sodium phosphate supplementation cause GI distress?

It is possible that high doses of sodium phosphate can cause
gastrointestinal distress. To minimize this side-effect, sodi-
um phosphate is best ingested in smaller doses spread in
three to four doses throughout the day, with food.

What about the association between serum phosphorous

and cardiovascular disease?
There is an association between serum phosphate and
cardiovascular disease, however this study showed that sup-
plementation does not raise serum levels and paradoxically
may even decrease them.

What I should know?

Sodium phosphate may be an effective supplement to
improve cycling performance in highly trained athletes, but
exact dosing strategies, performance benefits, and mecha-
nisms of action remain to be elucidated. ◆

Cyclists, endurance athletes, and sodium phosphate aficio-

nados (do they exist? maybe they should) can check out the
private Facebook ERD forum for more discussion on this
paper.

38
 On the whey to
getting lean: one
more round of
whey vs. soy
Whey Protein Supplementation
Preserves Postprandial
Myofibrillar Protein Synthesis
during Short-Term Energy
Restriction in Overweight and
Obese Adults
Introduction
Stuart Phillips’ group at McMaster University has been conducting research on
all things protein for close to two decades, with particular emphasis on how
protein feeding affects protein synthesis at the whole-body level, as well as specif-
ically in the muscle tissue (obtained from muscle biopsy specimens).

Measuring protein synthesis (or degradation) in muscle is the best biomarker of

ongoing muscle anabolism or catabolism that exists. Protein synthesis during
weight loss is important since keeping muscle on helps regulate blood
sugar and keeps metabolism higher. Obtaining information about
fat or protein turnover at a particular point in time can tell us
how well the body is conserving lean mass during weight loss,
which is what this study intended to do.

39
Who and what was studied? which is the major factor protein’s ability to stimulate pro-
tein synthesis.
The purpose of this study was to examine the effects of whey
and soy supplementation on protein synthesis and bio-
markers of lipolysis during weight loss, with carbohydrate
Figure 1: Grams of amino acids per 100
supplementation as the control. Even though body compo- grams of whey isolate or soy isolate
sition was measured during this study, the effect of whey
or soy protein supplementation on body composition was Essential Whey Protein Soy Protein
Amino Acid Isolate Isolate
not a major endpoint of the study. This was likely due to the
Isoleucine 6.1 4.9
relatively short duration of the study, as small body com- Leucine 12.2 8.2
position changes would not be detectable even with DXA Lysine 10.2 6.3
Methionine 3.3 1.3
(Dual X-ray Absorptiometry) measurements, which the
Phenylalanine 3.0 5.2
study uses and is considered the gold standard for body fat Threonine 6.8 3.8
measurement. Previous research has shown the error from Tryptophan 1.8 1.3
DXA to be around 1.2%, which would eclipse the relatively Valine 5.9 5.0
Total BCAAs 24.2 18.1
minor body composition changes from this study.
Total EAAs 49.2 36.0

There have been many studies comparing the effects of

In order to examine the effects of soy or whey on protein
soy and whey protein in stimulating protein synthesis and
synthesis during weight loss, 40 healthy, overweight and
affecting body composition in a range of conditions, but few
obese participants (19 males and 21 females) were includ-
that have looked specifically into the effect on protein syn-
ed and randomized to to receive either a twice daily whey
thesis and lipolysis during weight loss.
supplement (27 grams per supplement), soy protein supple-
ment (26 grams per supplement) or carbohydrate (CHO)
The studies that have previously compared whey to soy in
controls (25 grams per supplement), resulting in a total
their ability to stimu-
daily protein intake
late protein synthesis
of 1.3 grams per kilo-

  [...] research has

and stimulate gross
gram of bodyweight
muscle hypertrophy
in the protein-supple-
in other contexts have
generally found whey
to be superior to soy.
shown the error from mented groups and 0.7
grams per kilogram

As seen in Figure 1,
this is probably due to
DXA to be around 1.2%, of bodyweight in the
CHO group. All food

which would eclipse the

was provided by the
whey having a more
research team in the
complete amino acid
form of prepackaged
profile (most soy is
very low in two of relatively minor body meals, to increase
compliance.

composition changes
the essential amino
acids, methionine and
During the study, all
lysine) and especially
due to whey’s high-
er leucine content,
from this study. participants were first
put on a three-day

40
maintenance diet (given maintenance calories and 1 gram
per kilogram of body weight of protein), subjected to the
first test day, and then put on 14 days of a hypocaloric diet,
followed by the second test day. The purpose of the main-
tenance diet was to equilibrate the subjects and put them
on even ground, metabolically speaking. The hypocaloric
diet was calculated to provide an energy deficit of 750 kcals
per day, which should result in a weight loss of one to three
kilograms over the course of 14 days. On both of the test
days, the subjects received stable isotope infusions with
isotope-labelled phenylalanine and glycerol, were DXA
scanned, and had blood and muscle samples taken.
Body composition (total, fat and lean mass) was obtained
from DXA scans. Muscle protein synthesis was calculated
from tissue enrichment of stable isotope labelled phenyl-
alanine in muscle samples. Lipolysis (fat burning) was
calculated using the appearance of the isotope-labelled glyc-
erol in the blood. Glucose and insulin levels were obtained
from blood samples.
Overweight and obese participants were ran-
domized to receive either soy or whey protein,
or a carbohydrate control, and put on a hypoca-
loric diet. Body composition measurements and
muscle protein synthesis were measured and com-
pared between groups.
What were the findings?
The researchers found no significant difference in body
composition between the groups. This was actual-
ly the expectation for a study of this duration and diet
composition.
It has been shown numerous times that whey is a so-called
“fast” protein, which means that the amino acids in it are
absorbed faster than other proteins with a similar amino
acid content. In this study we see this yet again, as release
of leucine, essential amino acids and total amino acids were
significantly higher after ingestion of whey protein than
after soy. Soy still outperformed the CHO control supple-
ment. In fact, the plasma leucine release (measured by the
area under the curve [AUC]) from whey was almost three
times higher than that for soy, whereas the essential and
total amino acid AUC’s were approximately twice as high.
Not surprisingly, there were also differences in insulin and
glucose concentrations following ingestion of the supple-
ment between protein groups and the CHO control group,
both before and after the dietary intervention. Ingesting the
glucose control supplement resulted in higher glucose and
insulin levels than ingesting either protein supplement, with
no difference between soy and whey protein supplementa-
tion groups. As amino acids are absorbed more rapidly into
the bloodstream from whey protein than soy protein, it is
plausible to expect higher blood glucose and insulin levels
after whey supplementation.
There was also a significant difference between the amount
of circulating glycerol between the protein groups and the
CHO control group. Glycerol was lower in the CHO group
than in either protein group. This indicates that lipolysis is
suppressed following CHO ingestion relative to both of the
protein groups. In the post-prandial
condition, the amount of glycerol
in the blood is a measure of the
rate of ongoing lipolysis. Fat is
stored as triglycerides, and
when fat stores
receive neu-
roendocrine
signals to
mobilize
fat, tri-
41
   [...] this study actually does not prove
that whey is superior for maintaining
lean mass during weight loss. [...] whey
probably is better, but that cannot be
determined from these results because
this study lacks statistical power to find
such differences.
glycerides are hydrolyzed into fatty acids and glycerol, both
of which are released to the circulation. The fatty acids Whey protein was absorbed more quickly than
are metabolized in most of the body, whereas most of the soy protein, and stimulated muscle protein syn-
glycerol is used for gluconeogenesis. The most likely expla- thesis by roughly two times the amount that soy
nation for the lower glycerol levels observed following CHO supplementation did. However, no differences in
ingestion is probably that insulin strongly inhibits lipolysis.
overall body composition was observed between
Since it was shown that the CHO supplement resulted in
the groups.
higher insulin, this further reinforces this hypothesis.

The main finding of the study was the effect of the protein
supplements on protein synthesis in the myofibrillar pro-
tein fraction of muscle (the so-called Fractional Protein
Synthesis, FPS), before and after the dietary intervention.
The big picture
The results regarding amino acid, glucose, insulin, and glyc-
erol concentrations were in line with numerous previous
observations.

The researchers measured FPS before and after protein

ingestion, as well as before and after the dietary interven-
tion. The baseline FPS was around 0.02-0.03%/hour, but
upon stimulation it increased to 0.06-0.07%/hour for whey
and 0.03-0.04% for soy. When the FPS was expressed as the
change in FPS from before to after supplement ingestion,
soy scored higher than the CHO control supplement and
whey scored higher than both. This means that whey stim-
ulates the protein synthesis in muscle more efficiently than
soy protein does.
When it comes to the findings for body composition, this
study actually does not prove that whey is superior for
maintaining lean mass during weight loss. There’s a fair
chance that whey probably is better, but that cannot be
determined from these results because this study lacks sta-
tistical power to find such differences.
In order for differences between groups to manifest as sta-
tistically significant, a sufficient number of subjects must

42
be enrolled. If the between-group difference we are looking
at is small relative to the variation in the observations, this
calls for more subjects. The study duration was short (14
days) and therefore the changes in lean and fat mass are
small compared to the margin of error using a DXA scan-
ner, meaning that if any difference did exist between groups,
it would have taken more participants or a longer study
duration for significant changes to manifest.
If you know the variation in the measurement you are doing,
you can actually calculate
the number of subjects
needed to detect a group
difference of a given size.
  [...] whey stimulates
This is called a power
analysis. The researchers protein synthesis in
myofibrillar proteins
actually describe that
they knew they had inad-
equate power to detect
between-group differenc-
es, which was acceptable
much more efficiently
as this was not part of
the primary objective for
than soy protein
the study. Alternatively,
this may be a real and
supplementation.
valid finding. It is possi-
ble that although one protein source may be more effective
than another in a short window of time for measuring FPS,
overall dietary intake may be most important for producing
long-term adaptations in body composition.
What the study did show, however, was that whey stimu-
lates protein synthesis in myofibrillar proteins much more
efficiently than soy protein supplementation. How this
translates into lean mass sparing during weight loss is very
hard to derive for numerous reasons. FPS is normally used
as a surrogate biomarker for a snapshot of hypertrophy pro-
cesses. However, hypertrophy or atrophy is the result of net
protein synthesis or degradation, which is again the prod-
uct of gross protein synthesis and net protein degradation.
Therefore, it should be apparent that modulation of protein
degradation is just as important as modulation of protein
synthesis. This is supported by the results of some studies,
which have shown that different protein sources may have
different influences on protein synthesis and degradation.
The reason that protein synthesis is more frequently report-
ed is that the technology for measuring protein synthesis
is much better than the technology available for measuring
protein degradation. All of these factors make it difficult
to conclusively state whey’s superiority over soy based on
this study. The protein
synthesis measurement
used in this study is a
surrogate for the effect
on muscle mass. But
changes in muscle
mass can be measured
directly with something
like DXA, albeit this
requires a more chal-
lenging study setup,
meaning more subjects
for longer time and a
higher study cost. Of
course, studies can
always be better or
more detailed, so maybe this study will be the springboard
for the next, more detailed study.
Lastly, funding for this study was provided by the Dairy
Research Institute through the Whey Protein Research
Consortium. However, the reported findings are in
agreement with the literature and with other studies not
supported by dairy farmer’s associations.
Frequently asked questions
How much daily protein synthesis and degradation happens
normally, and under weight loss or muscle gain conditions?
Baseline fasted FPS in this study was around 0.02-0.03%/
hour, whereas the fed FPS was two to three times higher.
Based upon these numbers, and MPS and FPS data record-
43
ed elsewhere in the literature, we
can assume an average total pro-
tein synthesis of 1.0-1.5%/day.

This means that 1.0-1.5% of the

protein of the muscle protein
is built up and broken down per
day at muscle mass equilibrium.
The amount of muscle made out of
protein is constant, at around 20% of
the wet weight of muscle. As a normal
adult (60-80 kilograms / 130-175 pounds)
carries 30-40% of his or her body weight as
muscle mass, this corresponds to 200-400 grams
of muscle (or 40-80 grams of muscle protein) that is
built up and broken down, daily.

In the context of changes in muscle mass during sustained

weight loss or resistance training, the net changes in mus-
cle mass are generally on the order of 10-50 grams per day.
These can be bigger at the onset of weight loss or weight
gain, but it still underscores that much more protein is
being built up and broken down on a daily basis, than that
which is ultimately gained or lost as hypertrophy or atrophy.
Has there been research connecting muscle protein synthesis
to the end result of muscle hypertrophy?
A previous paper from Phillips’ research lab compared the
protein synthesis measured following a resistance exercise
bout in a part of a chronic resistance training study with the
gross hypertrophy measured after 16 weeks of the resistance
exercise program. The researchers found no correlation
between the two measures. Even though that was in a train-
ing study, the result underscores that maybe using protein
synthesis is not as good of a biomarker as we sometimes
would like to think.
What about molecular biology markers of muscle gain and
loss -- do they have similar findings as muscle protein syn-
thesis?
It is also worth noting that there is a disconnect between
protein synthesis measurements and several molecular
biology biomarkers, such as insulin signaling through the
Akt/mTOR signaling cascade or FOXO transcription factor
expression. In this case, however, this is more likely caused
by the poor reliability of the molecular biology biomarkers,
as they can display anabolic signaling even in grossly cata-
bolic cachectic subjects.
What should I know?
Ingestion of whey protein stimulates myofibrillar protein
synthesis during weight loss more efficiently than soy pro-
tein and naturally better than carbohydrate. This indicates
that protein obtained from whey may more efficiently spare
muscle mass than protein from soy, but the actual impact is
outside the capabilities of this study to assess. ◆
To dig into the science of protein synthesis a bit more, head
over to the Facebook ERD forum.

44
 It’s (not) all in your
head: how sodium
intake affects
headaches
Effects of dietary sodium and the DASH
diet on the occurrence of headaches:
results from randomised multicentre DASH-
Sodium clinical trial

Introduction For example, it was once believed that monosodium gluta-

mate (MSG) could cause headaches. This belief was largely
Though headaches are very common, very little is known
the result of a doctor who wrote a letter to the New England
about what causes them. Obviously, certain dietary choices
Journal of Medicine on “Chinese restaurant syndrome,” his
(namely too much alcohol or not enough water) can cause
term for the physical effects of his own overindulgences.
headaches, so it makes sense that other foods could also
The link between MSG and headache has not been proven
cause headaches. However, the evidence for this claim is
in rigorous studies, and the evidence is covered on Examine.
spotty at best.
com. Yet that single letter was enough to make MSG the
most researched dietary factor related to headaches.

45
Headaches have been linked to other health and lifestyle
factors too. One of the most common factors is hyperten-
sion. Because hypertension is associated with headaches,
and blood pressure levels have a well-known dietary com-
ponent, it is reasonable to believe that dietary factors that
influence hypertension may also influence the occurrence
of headaches. Therefore, the authors of this study set out to
examine the links between headache occurrence and two
dietary factors known to reduce hypertension: the Dietary
Approaches to Stop Hypertension (DASH) diet and reduc-
ing sodium intake.
needs and compliance with the study protocol.
Compliance is critical for a detailed trial like this, so par-
ticipants were required to eat at least one meal at the study
site, five days per week. All of their other food was provid-
ed when they came for their on-site meal. The participants
were asked to record any uneaten food as well.
After the run-in period, participants were randomized to
either the DASH diet or a “typical American diet.” Each
participant stayed on the same diet during the entire study.
They were then further randomized to one of three different

Who and what was studied sodium levels (low, medium, or high), with the highest level
being equivalent to the “typical” American sodium intake.
This study on headaches was part of a much larger study of
Every 30 days, each participant was switched to a different
the DASH diet and reduced sodium intake in hypertensive
sodium level after a seven-day “washout” period, until each
or prehypertensive individuals. The original goal of that
participant had received all three sodium levels.
study was to analyze whether the DASH diet and reduced
sodium levels have synergistic effects on blood pressure.
During the last week on each sodium intake level (at the
This study is a secondary analysis of 390 participants from
end of every month), the participants were tested for blood
the original trial, being the roughly 95% of participants who
pressure, body weight, and via urinalysis, and were sur-
completed side-effect questionnaires.
veyed for a variety of adverse events such as headaches
and fatigue. The urinalysis served a dual purpose of both
Since the original study was a major trial designed to test
assessing metabolic factors and ensuring compliance (by
interactions between multiple dietary interventions, it was
measuring sodium secretion). This study design is a great
extremely well controlled, and that feature carried over into
example of a well-designed randomized crossover study
this study. Participants had three separate baseline visits
because it has strict adherence policies, multiple assess-
separated by at least seven days, and during that time (about
ments of compliance, and blinded data collection staff.
two weeks), they were fed a run-in diet to assess caloric

Are secondary analyses second-class

citizens in the research world?
Secondary analyses are often treated the same as analyses from the original study, at least in the
media. But they probably shouldn’t be. The original study that this was based on wasn't designed to
examine the headache issue, so it's not quite as convincing as a study that was.

The reason has to do with statistics. When you state a primary hypothesis before launching the study,
conclusions are more statistically reliable than doing post hoc hypothesis testing. This is because
you can fish around for significant results and make a paper out of whatever you find (which may or
may not be due to chance, despite a significant p-value), as opposed to being forced to focus on your
main hypothesis no matter what the outcome was when you state it a priori.

46
Study participants were split into two groups: eating a diet designed
to reduce hypertension, or a typical American diet. Then, each
participant spent a month eating at each of three different sodium
levels: low, medium, or high. Researchers measured blood pressure
Figure 1: Incidence of
and body weight, as well as tracked when headaches occurred. severe headaches by
diet and sodium intake

What were the findings?

The high sodium diets (whether “typical” or DASH) had a significantly greater
incidence of headaches compared to the low sodium diets.

Going from low to intermediate to high sodium levels steadily increased headache
odds in both the DASH diet (36% to 38% to 43%) and control diets (39% to 41% to
47%). However, the difference in headache risk between the intermediate sodium
intake group and the lowest and highest sodium intake groups was not statistical-
ly significant. This may be related to the fact that this is a secondary analysis of a
study not specifically designed to analyze headache occurrence. Headaches were
were only measured once in each study period, and the study didn’t collect data on
previous headache prevalence, among other design limitations.

Interestingly, the association between headache occurrence and sodium intake

persisted regardless of hypertension status or blood pressure. As one might
expect in a trial of hypertensive individuals who were fed the DASH diet, some
participants experienced a decrease in blood pressure below hypertensive levels.
However, even in these individuals, the correlation between headache risk and
sodium intake remained.

Perhaps most surprisingly, there was no overall difference in headache occur-

rence between the “healthy” DASH diet and the “typical” American diet. There
was however a difference between DASH dieters eating low sodium (36% had
headaches) and control dieters eating high sodium (47% had headaches).

The study participants also classified their headaches as mild, moderate, or severe.
Most headaches were classified as mild, and higher sodium intakes generally were
associated with a bit higher (although not statistically significant) mild-to-moder-
ate headache rates. Despite the small number of severe headaches, Figure 1 shows
that high sodium seemed to have an association with severe headache incidence.
This association trended toward significance in both diet groups.

47
What does this tell us?
It is well-known that sodium affects hypertension and that
hypertension affects headaches, and this study indicates
that sodium affects headaches in hypertensive individuals.
However, diet and hypertension status did not play a role
in the interaction between sodium intake headaches, which
suggests that sodium intake and headache risk may be
linked even in people who are not hypertensive.
This is supported by the data that show that hyperten-
sion status was not correlated with the effects of sodium
on headache risk, but it is also worth mentioning that the
hypertensive participants had significantly more headaches
than the people with normal blood pressure. So there may
be more to the interactions between hypertension, sodi-
um, and headaches than can be revealed by the design of
this study. It’s possible that sodium may impact headaches
through mechanisms unrelated to blood pressure.
It is important to remember that this trial, no matter how
well-designed, was not originally intended to study the
effects of sodium intake on headache. Therefore, these analy-
ses are based only on reports of headaches while on the study
protocol. Headache frequency was not assessed at baseline,
which may be one of the reasons that this study’s findings are
only significant for the high and low sodium intake groups.
This is an important factor to consider when comparing
these data to other studies on headache occurrence.
  Perhaps most surprisingly, there
was no overall difference in headache
occurrence between the “healthy” DASH
diet and the “typical” American diet.
The Big Picture
The correlation between hypertension and headaches is
well-established. Similarly, the correlation between dietary
factors and hypertension is well-known, in part thanks to the
“grandfather” of this and many other dietary trials: the DASH
study. On the opposite side of these famous and well-de-
signed studies is the largest body of “evidence” for dietary
factors that contribute to headache: a bunch of studies that
refute anecdotal evidence linking headache to MSG con-
sumption. Some believe that diets high in fruits and veggies
are a panacea of sorts, and can curb pretty much any malady
including headache. This study doesn’t support that notion.
This study does show that sodium intake and headache
are probably correlated, but it’s still not known how. The
simplest theory is that headaches are related to sodium’s
effect on blood pressure. This study provides relatively weak
evidence, since it shows that sodium intake and headache
correlate regardless of blood pressure levels, diet, or hyper-
tensive status. We’re left with a situation that’s as promising
as it is frustrating. It’s relatively certain that sodium intake
(or intake of sodium-rich foods) affects headache occur-
rence, but there’s no good explanation as to how that
happens. Regardless, this study still makes the case for lower
sodium consumption than is typical in modern diets, if one
is trying to reduce headaches.
However, other merits of reducing sodium intake have been
heavily debated recently. A recent study in the Journal of the
48
American Medical Association (JAMA) that found no associ- Heart Association recommends <2.4 grams sodium and
ation between sodium intake and mortality or cardiovascular says 1.5 is even better. The US Centers for Disease Control
disease. Though headaches are not the same as death, the recommendations are similar to those of the AHA.
differences between the studies are still worth mentioning.
Notably, the JAMA study analyzed data from a much longer Generally speaking, if you have cardiovascular disease or
time frame in a much larger cohort of patients, but it did so hypertension, less than 1.5 grams of sodium per day is a
using self-reported data. In contrast, this study followed a
smaller group of patients (but still sizeable by trial standards)
with much more rigorous requirements and assessments.
  It’s relatively
The World Health Organization (WHO) and other groups
already recommend a diet with significantly lower sodium certain that
than the average American consumes, due to the interaction
between sodium and hypertension. A reduction of headache sodium intake
(or intake of
occurrence may be an added benefit. The typical American
consumes ~182.7 mmol sodium per day, which is similar to
the 150 mmol received by the high sodium intake group in
this study. The current WHO recommendation is 87 mmol/
day. However, it should be noted that the WHO recom-
sodium-rich
mendation is still higher than the intake of the low sodium
group in this study (50 mmol). So although there may be a
foods) affects
general consensus of “eat less sodium,” there is not nearly
enough evidence to offer a firm recommendation beyond a
headache
general upper limit. And that is besides the recent contro-
versy about whether sodium reduction actually has benefits, occurrence, but
there’s no good
which could turn decades-old diet advice on its head.

For headache sufferers, this study provides a tantalizing bit

of well-controlled evidence, especially since effective pre-
vention for some people can seem quite elusive. But don’t
explanation
equate the results with causation -- it’s still possible that
high-sodium foods eaten in the study could cause headache
as to how that
independent of their sodium content. The data on head-
aches resulting from foods and nutrients is unfortunately
happens.
quite sparse.

Frequently Asked Questions good goal. If you’re otherwise healthy, you can probably get
away with eating up to 2.4 grams, but that is based on gov-
There are many different recommendations for sodium
ernmental and organizational guidelines, which may change
intake (WHO, AHA, etc). Which should I follow?
as evidence is changing.
That’s hard to answer definitively. The WHO recommends
<2 grams sodium (so <5 grams/day salt). The American

49
It’s important to note, however, that none of these recom-
mendations are based on headaches at all. They are strictly
related to the research on interactions between hyperten-
sion and sodium.
Is this study sufficient to recommend or adopt a reduced
sodium diet for people who suffer headaches?
It’s impossible to make a definitive statement about recom-
mendations based on one or even a few studies, no matter
how well-designed. However, if you struggle with head-
aches, reducing sodium consumption may be worth a try
because it’s a relatively easy intervention without any known
side effects and multiple potential benefits.
Is there any good reason to increase sodium intake?
It is often recommended that individuals engaged in very
strenuous activity (for example, ultramarathon runners)
should consume something with added sodium to prevent
hyponatremia, but there isn’t that much evidence to support
this assertion, and it may be more important to limit exces-
sive fluid intake.
Oral rehydration therapy (for people with major GI illness-
es or severely dehydrated individuals who need to recover
electrolyte levels) has sodium as one of its major com-
ponents, but the effects of that therapy are related to the
interplay between glucose, sodium, and water absorption in
the small intestine rather than sodium alone. Furthermore,
if a person experiences a GI illness that severe, they are
likely already receiving medical care that includes a com-
prehensive rehydration protocol. Individuals who engage in
extreme weight cuts for sport performance may consume
extra sodium to rehydrate, but it is likely more effective
when combined with glucose.
Lastly, those with low blood pressure can also find salt to be
useful when facing hypotensive episodes.
MSG has sodium in it. Could that possibly explain the
effects observed in some individuals consuming MSG?
Most MSG health effect reports are anecdotal. The effects of
MSG may be overstated, as blinded trials haven’t identified
any adverse events associated with MSG consumption, even
in individuals who self-identify as “sensitive” to MSG.
In fact, in one study, individuals who claimed to be sensitive
to MSG were repeatedly randomized in four separate trials,
and only two out of 130 responded appropriately to their
treatment in all four trials. Furthermore, regularly con-
suming enough MSG to notably affect sodium intake levels
would be a substantial feat because a gram of pure MSG
equates to only 150 mg of sodium. It would take over 13
grams of MSG to reach the maximum intake recommended
by the WHO (2 grams).
What should I know?
This study was very well-designed and had a rather strong
conclusion. It showed that sodium intake likely influences
the prevalence of headaches irrespective of other dietary or
demographic parameters. However, it’s important to note
that the effect size was not large, which implies that reduced
sodium consumption may reduce the frequency of head-
aches, but it’s rarely the sole cause and unlikely to be the
sole solution.
However, most people can stand to reduce intake of junk
foods that are often high in salt, and it’s unlikely to cause any
negative effects, except possibly reducing your consumption
of delicious and addictive snacks you ate as a child. ◆
If all this analysis is giving you a headache, take a load off
on the Facebook ERD forum.
50
 Diets, fast and slow
The effect of rate of weight loss on
long-term weight management: a
randomised controlled trial
Introduction
Many people say that dieting is hard, and that keeping the lost weight off is even harder. But
we don’t need to take their word for it, the science of weight loss backs up their claims. One
review, containing evidence gathered during most of the 20th century, found that only about
15% of obese people were successful in keeping a significant amount of weight off long-term
through dieting!

Why is dieting so hard? Maybe it’s because the way we go about dieting isn’t quite right.
Government agencies in the U.S., U.K., and Australia all recommend an initial, gradual weight
loss strategy of one to two pounds a week. Gradual weight loss is recommended for several
reasons. It allows more time for people to form the habit of eating fewer calories. Also, crash
diets with very low daily caloric intake can sometimes lead to nutritional deficiencies. Plus, it’s
claimed that crash diets lead to larger rebounds after the diet is over, which makes sense con-
sidering that very low calorie diets have been seen to lower basal metabolic rate.

51
But is this true? There is certainly some evidence to suggest
that it is. For instance, one study found that counselling
overweight and obese women to make small changes in
their diet with the goal of 1800 calories per day succeeded
in slow, steady and sustainable weight loss that persisted
even after a year. A control group that provided counselling
encouraging a 1200 calorie/day diet ended up with regain-
ing weight after a year.
However, there is disagreement about whether or not slow
weight loss actually leads to better outcomes. One review of
the literature suggests that faster initial weight loss was asso-
ciated with better long-term weight maintenance, as long
as proper support is provided. A recent trial, which encour-
aged older, obese women to decrease daily caloric intake
to 1200 while increasing physical activity, found that the
people who achieved more rapid weight loss both lost more
weight and kept more weight off. However, this trial was not
randomized, so it’s hard to say whether the rapid weight loss
caused the increased success in weight maintenance or not.
Sustained rapid weight loss is a bit more common in clinical
settings. Protein-sparing modified fasts (PSMFs) are some-
times prescribed to very obese patients, and have calorie
intakes of 500-600 kcal per day. For these patients, PSMFs
can have better long-term weight loss than other approach-
es. While there is a
rebound, as with any
diet, adherence is fairly
high and people remain
  [...] only about 15%
of obese people were
motivated because they
can see the scale mov-
ing. These patients, as
they tend to be very
successful in keeping
heavy, also tend to face
substantial and often a significant amount
immediate health risks
if weight is not lost. of weight off long-term
The purpose of this through dieting!
study was to address
the relative lack of evidence comparing rapid and gradual
weight loss. By randomizing rapid versus slow weight loss,
the researchers hoped to answer whether or not the speed of
initial weight loss actually affects keeping the weight off.
Guidelines currently suggest that slow weight
loss is preferred to rapid weight loss, since rapid
weight loss may not lead to sustainable results
over time. The evidence, however, is not clear-cut
on this matter.
Who and what was studied?
Participants were recruited via advertisements in
Melbourne, Australia. To be included in the trial, the par-
ticipants had to be obese (defined as a BMI between 30 and
45) but otherwise healthy adults between 18 and 70 years
old. People who smoked, were on medications that could
affect weight, had any clinically significant disease (such as
diabetes), or who used weight-loss drugs or adhered to a
very low calorie diet three months prior to recruitment were
all excluded.
The 204 participants were studied in two different phases.
The goal of the first phase was to have participants lose 15%
of their weight, either
quickly or slowly. The
second phase took only
people who met this
goal, so they could
be observed to deter-
mine how well their
weight loss could be
maintained.
The first phase was per-
formed by randomly
assigning participants
to either a fast or slow
weight loss program.
52
Both programs had the aim of a 15% loss in weight using an
overall caloric deficit of 105,000 calories, but over different time
frames and using different methods. The fast program consist- Random samples,
ed of replacing three meals a day with Optifast, a commercial
meal replacement. The total caloric intake for this group varied
random assignment
between 450 - 800 calories a day, depending on the individual’s
This study was a randomized trial. But what
needs to reach their 15% weight loss goal over 12 weeks. This
exactly was randomized, and why does ran-
rate of weight loss is about equal to 1.5 kilograms or 3.3 pounds
domness matter in the first place? There are two
per week.
types of randomness that pertain to trials like
this: random sampling and random assignment.
The slow group’s diet consisted of a 400 - 500 calorie defi-
cit achieved by using one or two Optifast meal replacements,
Random sampling is when each member of a
alongside a diet concordant with the Australian Guide to
population has an equal chance of being select-
Healthy Eating from 1998 (which has since been updated to
ed for a study. It allows you generalize the result
allow for more protein), which suggests that 15% of daily cal-
of a study from the subset included in the study
ories come from protein, 25–30% from fat, and 55–60% from
to the entire population. Why can’t you do this
carbohydrates. The aim of the slow diet was for each participant
without a random sample? Because there may
to lose 15% of their weight over 36 weeks, which would average
be bias otherwise. For instance, if we put a poll
out to about 0.5 kilograms or 1.1 pounds per week.
up on Examine.com to ask people “Do you care
about the evidence behind abstracts?” we’d
Both groups were given the same educational material and
probably get a large number of “yes” responses.
assigned to biweekly meetings with a dietician, who provided
You Examine.com readers exceptionally curious,
feedback on their progress. Both groups were also encouraged
and definitely not a random and representative
to exercise at a mild or moderate intensity for at least 30 min-
sample.
utes a day. Study participants who succeeded in losing at least
12.5% of their weight in both groups moved on to the second
This study did not use a random sample of peo-
phase of the study.
ple, but instead enrolled people who answered
an ad or heard about the study through word of
The purpose of the second phase was to determine whether
mouth. Since this sample could be biased (for
the rate of weight loss affected the ability to maintain the loss-
instance, perhaps those who answered the ad
es. Each participant received an individualized plan, which
were very motivated to lose weight), it would be
again adhered to Australian Guide to Healthy Eating. Each
hard to generalize the results of this study to all
participant’s goal was to maintain their weight. They met with
people.
a dietician four and 12 weeks into phase 2, and every 12 weeks
thereafter, for 144 weeks. Their adherence to the diet was mon-
However, this study used another kind of ran-
itored via self-reported food intake logs. If a participant was
domness that is also important: they randomly
gaining weight, they were advised to change their diet to create a
assigned participants to groups. This helps to
400 - 500 calorie deficit.
infer causality in a study because it helps to get
rid of confounding factors.
In addition to weight, fasting ghrelin (a hormone that induces
hunger) and leptin (a hormone that induces satiety) blood levels

53
 What were the findings?
  One of the Almost 90% of participants completed phase 1 (the weight
loss phase) of the study, with significantly more people

biggest findings dropping out of the slow weight loss group (18% vs. 3%).
The biggest reason cited for dropout was difficulty of stick-

from this study ing to the diet.

is that the rate of As seen in Figure 1, more people in the fast weight loss
group achieved their goal of reducing their weight by 12.5%,

weight loss does

and the result held up whether all subjects were considered
or just those that completed phase 1. There was no signifi-
cant difference in changes in fat-free mass in either group.
not affect weight
Figure 1: Rate of weight loss
regain in the tested
population
and subjective hunger assessments were also measured at
the start of the study, the end of phase 1, and weeks 44 and
144 of phase 2. A subgroup of participants also had their
3-beta-hydroxybutyrate levels drawn, which is a measure of
ketosis. Activity was measured with a pedometer.

This trial was implemented in two phases. In

phase 1, obese adults were randomized into two
diets designed to reduce their weight by 15% over
either 12 weeks (the fast group) or 36 weeks (the
slow group). In phase 2, participants who suc-
ceeded in losing at least 12.5% of their weight
were put on a maintenance diet designed to main-
tain the weight loss. Researchers intended to
determine if rapid weight loss leads to rebound There were some biochemical differences between the
weight gain, and what role hunger plays in the groups in phase 1 as well. The fast weight loss group had
rebound process. significantly higher levels of 3-beta-hydroxybutyrate (indi-
cating ketosis) partway through phase 1, but there was
no difference between groups at the end. Leptin (which
promotes satiety) decreased throughout phase 1 for both

54
groups, but decreased significantly more in the fast weight Figure 2: Success rate during weight loss
loss group. Ghrelin (a hormone that induces hunger) rose
phase (12.5% or more weight loss)
equally for both groups throughout phase 1. Both groups
reported an increase in subjective hunger, although there
was no difference between the groups.

The fast weight loss group tended to walk about a thou-

sand steps more per day than the slow weight loss group,
even though they were both given the same exercise advice.
However, the amount of steps taken did not differ between
groups when only participants who succeeded in losing
12.5% of their weight were examined. People who walked
more were not more likely to meet their goal, as there was
no significant difference in number of steps taken per day
between people who met the weight loss goal and people
who didn’t.

Overall, 51 participants from the slow weight loss group and

76 from the rapid group moved on to phase 2 of the study in
order to determine if the weight loss could be maintained.

There was no significant difference in the dropout rate

between groups in phase 2 of the study. All but six people
(one from the slow group and five from the fast group)
started regaining weight, and so were put on a 400 - 500
calorie per day deficit, as per the experimental protocol. As
seen in Figure 2, during phase 2, no significant difference
in weight gain was seen between groups; both regained
over 75% of the weight they originally lost on average. This
put both groups at a total of about 5% weight loss over the
course of the study. The pedometer measurements also
didn’t differ between groups.
Leptin rose at a similar rate in both groups in phase 2, but
participants who regained less than 25% of the weight lost
in phase 1 had much lower leptin levels than those who
gained more than 75% of it back. Ghrelin levels decreased
a little during phase 2, but still remained elevated, with no
significant difference between groups. However, subjective
hunger was higher in the group that lost weight quickly
than in those who lost it slowly.
There were also some differences in adverse events between
the groups. During phase 1, one person developed acute

  Also, crash diets with very low daily

caloric intake can sometimes lead to
nutritional deficiencies.
55
 cholecystitis, a condition known to occur with rapid weight changes, where the bile
duct of the gallbladder is blocked, which leads to inflammation. The authors of the
study considered this to likely be caused by the rapid weight loss. Two other serious
adverse events occurred in the rapid weight loss group (cancer) that the authors do
not think was caused by the rapid weight loss. One person in the rapid weight loss
group also withdrew due to worsening depression, on which the authors did not
comment. There were no adverse events in the slower weight loss group.

The fast weight loss group had an easier time sticking to the diet
than the slow weight loss group. The fast weight loss group was also

  It’s quite more successful in losing at least 12.5% of their weight in phase 1 of
the study. Both groups gained back about 30 pounds during phase 2

apparent while trying to maintain their weight loss, with no significant differ-
ence between the two groups. Hormonal and subjective measures of

that the hunger rose throughout phase 1, and only dipped slightly in phase 2.

body fights What does the study really tell us?

losing One of the biggest findings from this study is that the rate of weight loss does not
affect weight regain in the tested population: obese nonsmokers, who don’t take

weight and
any weight loss drugs and who don’t have diabetes. This puts a dent in the idea
that rapid weight loss leads to greater weight gain down the road. It also tells us
that rapid weight loss can be even more successful than gradual weight loss, since
keeping it more people were able to stick to the rapid weight loss plan than the gradual one,
while also meeting their weight loss goal.

off. It also tells us something many of us already know: dieting is hard. The total
weight loss for both groups ended up being about 5%. Most of the weight lost
during the dieting phase of the study was regained during the maintenance phase,
although that phase was an incredible 144 weeks long, so it’s not like the weight
was regained right away. It’s also important to note that results reflect averages for
the sample -- some may have been wildly successful, while for others life events
may have gotten in the way, as they often do. In other words, there can be quite a
bit of human variation when trying rapid vs gradual weight loss.

It’s quite apparent that the body fights losing weight and keeping it off. Hunger
was observed to increase and persist in both groups throughout both phases,
boosted by a rise in ghrelin (a “hunger hormone”) that persisted for at least three
years (the length of this study), and which doesn’t level off much, even as weight
is regained. Leptin levels (a “satiety hormone”) also remained low in people who

56
kept weight off, but rose in people who put significant
weight back on, suggesting that the body is attempting to
get its weight back to where it started.
  Making more
Keep in mind, the people in this study had access to nutri-
tional counseling and were under medical supervision
choices drains
throughout the course of the study. The authors note that
because of this, the results of this study may actually be willpower, which
better than what could be expected in the real world, where
people are not under such strict supervision. This was also has been shown
one conclusion of a systematic review of weight loss studies,
and suggests that the methodology of rigorous clinical trials to affect food
like this may make their results less generalizable outside of
the context of a controlled study environment. The fact that
the study’s participants did not have diabetes and did not
choice.
smoke, and that a specific brand of meal replacement was
used, may further limit the generalizability of these results. meal replacements that use high-quality protein and are for-
tified with nutrients have a better safety record when used
But, since this was a randomized trial, a pretty strong con- under medical supervision. Early PSMF diets employed for
clusion can be drawn: fast weight loss does not necessarily rapid weight loss used fairly incomplete proteins like col-
cause fast regain, and can be a relatively effective method for lagen, along with very few micronutrients, and thus do not
weight loss for some people. apply to modern PSMFs.

This study found that losing weight quickly does
not lead to any more weight gain than losing
weight more slowly, and in fact can be more effec-
tive. However, it may be difficult to generalize
There are some aspects of the rapid weight loss plan that
may have helped the participants of this study be more
successful. First, participants in the fast group in this study
didn’t have to measure portions, as they completely relied
on meal replacements to lose weight, whereas people in the

these results to different populations outside of a slow group still ate some conventional meals. Measuring
portions to count calories isn’t always accurate and can be
clinical setting.
hard to learn, which may explain some of the fast group’s
success. Also, the use of meal replacements eliminates
variety in the diet, which helps satiety and leads to less food
The big picture eaten, and has been seen to increase diet compliance and
Despite the fact that some government agencies and “com-
weight loss in another study. The psychological phenome-
mon sense” dictates that losing weight more slowly is more
non of decision fatigue may also play a role here. Making
sustainable, this study showed that rapid weight loss can be
more choices drains willpower, which has been shown to
a more effective method of losing weight, and doesn’t neces-
affect food choice.
sarily lead to more weight gain. Very low calorie diets have
had a history of safety concerns, primarily caused by poor
While caution in extrapolating the results of this study to
nutrition. At such a low calorie intake, you’re unlikely to
other populations is warranted, this study is an important
get the nutrients you need to be healthy. However, modern

57
contribution to the science of dieting. According to the
authors, there were no randomized, controlled trials specifi-
cally studying the effects of the rate of weight loss on weight
maintenance. Due to the relative lack of evidence for the
widespread belief that rapid weight loss can lead to stron-
ger rebound weight gain, this belief has been questioned in
some recent papers. The current study adds to the evidence
supporting that, at least for some people, some of the time,
slow and steady does not always win the race.
This study suggests that while very low calorie
diets have been shown to be be potentially harm-
ful in the past, they may be a viable weight loss
option for obese people if done under proper
nutritional and medical supervision. There are
several behavioral reasons why modern meal
replacements may have contributed to the success
of the fast weight loss group in this study.
Frequently asked questions
How much of the weight loss came from muscle in this
study?
Fat-free mass was measured at the end of both phases
using bioelectrical impedance, and it did not significantly
decrease in either group. While bioelectrical impedance
is often considered a worthless, garbage, waste-of-time
measurement among those who are really into fitness, it’s
actually not a bad way to measure changes in a single indi-
vidual over time.
However, it is important to point out that muscle loss was
not measured throughout the weight loss phase. But while
we can’t really say too much about muscle loss from this
study, a large change in muscle loss would have likely shown
up in the group means.
The people in this study only lost 5% of their bodyweight by
the end of the study. That’s not much, right?
Modest weight loss can still have several benefits. Even a
5% drop in weight can help significantly lower the risks of
developing many of the chronic diseases associated with
obesity.
Keep in mind that the study participants regained weight
over three years. After one year, they had still retained more
than half of their initial weight loss. Maybe if the study
implemented a one-year anniversary kick-in-the-butt to
get participants back on track, the results might have been
different. But as it stands, the study protocol was the same
throughout the three years of maintenance, and that didn’t
really work out.
Can crash diets be harmful?
Yes, in some circumstances they can be. In this study, one
person suffered from an inflamed gallbladder, which was
attributed to the diet. The American Association of Clinical
Endocrinologists mention several other dangers, such as
uric acid buildup in the blood from very low calorie diets,
which can lead to severe complications, including kidney
damage. These kinds of diets, with calories under 1000 kcal
per day, should be done under medical supervision.
What should I know?
This study found that medically-supervised, rapid weight
loss in obese individuals leads to no more weight gain after-
ward than slower weight loss. This method also had a higher
success and compliance rate. This runs counter to current
recommendations, which suggest that rapid weight loss is
harder to achieve, and can lead to greater weight gain down
the road. ◆
Hungry for more? Whet your appetite for discussion in the
private ERD readers’ Facebook group.
58
 Is the glycemic
index actually
useful for making
food choices?
Effects of High vs Low Glycemic
Index of Dietary Carbohydrate
on Cardiovascular Disease Risk
Factors and Insulin Sensitivity
Introduction
Foods that contain similar amounts of carbohydrates may differ in the amount and
speed at which they raise blood glucose levels. This quality is referred to as the
“glycemic index” (GI), and is based upon the blood glucose response (total
area under the glucose curve) over two hours to 50 grams of its carbohy-
drate content, as compared with 50 grams of glucose.

The higher a GI value is, the more the food raises blood sugar in a
way similar to pure glucose. Lower GI carbs are those that digest
more slowly and don’t raise raise blood glucose to the same degree.
However, GI can also be a function of insulin response, and not
just a function of carbohydrate digestion or meal composition.

It has become common practice among healthcare professionals to

suggest an individual consume “low glycemic index” carbohydrates.
However the effects of GI on diabetes and cardiovascular disease (CVD)
risk factors have been equivocal in trials, as well as meta-analysis of
observational studies.

59
Further complicating matters is the fact that carbohydrates
are not eaten in isolation, but generally as part of mixed
meals containing fat, fiber, potassium, polyphenols, and
other nutrients that can affect health measures. It is still
uncertain what (if any) practical significance there is for
manipulating GI in the context of an otherwise healthy diet.
The DASH diet, or Dietary Approaches to Stop
Hypertension, is an eating plan recommended for people
with hypertension or prehypertension. The DASH diet has
been shown to reduce blood pressure, lower risk of coro-
nary heart disease, and improve bone mineral status. The
diet puts a big emphasis on increasing fruit and vegetable
consumption, controlling sodium intake, and reducing pro-
cessed food.
The objective of this randomized controlled crossover feed-
ing study was to determine the effects of GI and amount of
total dietary carbohydrate in the context of a DASH-type
diet, on various risk factors for diabetes and CVD.
Over half (52%) of the participants were women and
51% were black. The authors oversampled black partici-
pants because of their “disproportionate burden of insulin
resistance and other risk factors that result in high rates
of diabetes and cardiovascular disease.” Signs of metabol-
ic dysfunction were present in many of the participants,
including hypertension (26%), obesity (BMI ≥30) (56%),
LDL cholesterol at or above 130 mg/dL (68%), triglycerides
at or above 150 mg/dL (17%), and elevated (≥100 mg/dL)
fasting blood glucose (30%).
Four possible diets were provided based on the DASH-diet
guidelines. Participants were provided all of their meals,
snacks, and calorie-containing beverages throughout the
study. The intake of key factors such as total carbohydrates,
fiber, fatty acids, potassium, and sodium was controlled for.
While GI values do not take mixed meals into account, the
meals used in the study for high– and low– glycemic index
were put together using similar types of foods according to
their values (for instance, white bread versus whole wheat
bread). The four diets were:

Who and what was studied? 1. High–glycemic index (65% on the glucose scale),
A total of 163 overweight adults with above-normal blood
high-carbohydrate (58% of total energy)
pressure completed this trial, which was performed at Johns
2. Low–glycemic index (40% on the glucose scale),
Hopkins Medical Institution (Baltimore, MD) and Brigham
high-carbohydrate
and Women’s Hospital (Boston, MA). Participants were
3. High–glycemic index, low-carbohydrate (40% of total
at least 30 years old, had elevated blood pressure (120-159
energy)
mmHg systolic and 70-99 mmHg diastolic), and were classi-
4. Low–glycemic index, low-carbohydrate
fied as overweight or obese by having a BMI of 25 or higher.

A sample guide to the DASH diet

Type of Food Examples Number of Servings/day
Whole wheat bread,
Grains 6-12
rice, quinoa
Fruits Apples, bananas, oranges 4-6
Vegetables Spinach, tomatoes, carrots 4-6
Low fat dairy Milk, yogurt, cheese 2-4
Lean meats/seafood Beef, chicken, salmon 1.5-2.5
Nuts/seeds/legumes Almonds, cashews, beans 3-6/week
Fats/sweets Avocado, olive oil 2-4

60
   It is still uncertain what (if any) practical
significance there is for manipulating GI in
the context of an otherwise healthy diet.
Each participant completed at least two of the study diets for
five weeks each, with a two-week break consuming self-se-
What were the findings?
Figure 1 shows the primary outcomes at baseline and after
lected foods separating the diet interventions. Body weight
each diet.
was maintained by adjusting calorie intake, and participants
recorded any additional foods as well as foods they did not
eat. Maintaining body weight was important because simply Figure 1: Primary Outcomes at Baseline
losing weight can independently improve CVD risk factors. and at the End of Feeding on Each Diet
Despite these efforts, participants still lost an average of 1
kg from baseline until the end of each diet period. However
there were no differences between groups. The main out-
come measures were insulin sensitivity, LDL-cholesterol,
HDL-cholesterol, triglycerides, and systolic blood pressure.

Daily Low carb, Low carb, High carb, High carb,

intakes high GI low GI high GI low GI
Kcal 2011 1993 2011 1998
Protein (%) 23 23 16 16
Carbs (%) 41 40 58 57
Fat (%) 37 37 27 27
Fiber (g) 29 33 32 37
Potassium
3949 4026 39663 4103
(mg)
Glycemic
112 64 172 104
Load
Area under the glucose curve (AUC) – this is a way to mea-
sure how much and for how long a person’s blood sugar is
elevated. On the high-carb diet, eating ‘low GI’ reduced the
Study participants consumed at least two of the
12-hour AUC by 17% compared with high GI. However,
four available diets during the course of the study. on a low-carb diet there was no difference in glucose AUC
The different diets contained either low or high between high or low GI diets.
carbohydrate meals, using either high or low gly-
cemic index foods. Researchers measured insulin The low-carb, high GI diet reduced AUC by 19% compared
sensitivity, LDL-cholesterol, HDL-cholesterol, with high-carb, high GI, but there were no differences
triglycerides, and systolic blood pressure. between high or low-carbs diets with low GI. Finally, the
AUC was reduced by 20% in low-carb, low GI compared

61
with high-carb, high GI. To summarize, either low-carb or lowered with both low-carb diets, and increased on both
low GI will lower blood glucose through a 12-hour period, high-carb diets.
but choosing both will not provide additional benefits.
To summarize:
Insulin sensitivity - On the high-carb diet, the low GI diet
unexpectedly reduced insulin sensitivity by 20% compared Low-GI, Low-GI, Low-GI,
low-carb diet high-carb diet low-carb diet
with the high GI diet. There was no difference in insulin vs vs vs
sensitivity on the low-carb diet between high and low GI, High-GI, High-GI, High-GI,
high-carb diet high-carb diet low-carb diet
but another unexpected result was that the low GI diet
Insulin sensitivity
increased fasting glucose levels compared with high GI. Triglyceride levels decreased by 20% Triglyceride levels
Insulin sensitivity was not different between high and low- decreased by 23% LDL-cholesterol decreased by 5%
increased by 6%
carb diets (at either high or low GI).

Lipids and Blood pressure – LDL-cholesterol increased by What does the study really
6% on a high-carb, low GI diet compared with high-carb,
high GI. Glycemic index had no effect on HDL cholesterol
tell us?
In line with the prevailing opinion in the medical com-
level, systolic blood pressure, or diastolic blood pres-
munity, the authors of this study expected that consuming
sure. However, a low-carb compared with high-carb diet
foods with a lower GI would lead to improvements in insu-
increased HDL-cholesterol by 4% at high GI level.
lin sensitivity and CVD risk factors. Unexpectedly, the low
GI, high-carb diet compared with the high GI, high-carb
Regardless of GI value, low-carb decreased diastolic blood
diet actually decreased insulin sensitivity and increased
pressure by 1 mm HG compared with high-carb diets. On
LDL-cholesterol and LDL apo-B levels.
the low-carb diet, plasma triglycerides decreased by 5%
with low GI compared with high GI. When comparing the
It has been well established that a DASH-type diet can suc-
low-carb diet to high-carb, plasma triglycerides were lower
cessfully lower blood pressure, and the same authors have
by 18% and 20% with high and low GI, respectively.
previously shown that a lower-carb DASH diet can reduce
plasma triglycerides, VLDL levels, and diastolic blood pres-
Consuming the low GI, low-carb diet lowered triglycerides
sure. Varying the glycemic index of the carbohydrates in the
by 23% compared to the high GI, high-carb diet, but did
DASH-type of diet had yet to be studied. This study sug-
not affect insulin sensitivity, systolic blood pressure, LDL-
gests that in the context of a DASH diet, low GI foods do
cholesterol, or HDL-cholesterol. There were no additive
not improve CVD risk factors and may actually have some
effects (of carb level and GI) on any of the outcome mea-
detrimental effects compared with higher GI foods.
sures. Additional analysis showed the effects of the diets
were in the same among men/women, race (black and non-
On a high-carb diet (58% of daily energy intake), choosing
black), presence of metabolic syndrome, and BMI.
foods with a lower GI had no effect on levels of HDL cho-
lesterol, triglycerides, or blood pressure, and led to small
Changes From Baseline. All diets were successful in lower-
but unfavorable changes in LDL-cholesterol and insulin
ing blood pressure (systolic by 7-9 mm Hg and diastolic by
sensitivity. On a low-carb diet (40% of daily energy intake),
4-6 mm Hg). LDL-cholesterol was lowered with the high-
choosing foods with a lower GI had no effect on levels of
carb, high GI diet and both low-carb diets by 9-10% and
HDL-cholesterol, LDL-cholesterol, insulin sensitivity or
with the high-carb, low GI diet by 6%. Triglycerides were

62
blood pressure, but did lead to a small (5%) but favorable
decrease in plasma triglycerides. When compared to a high-
carb, high GI diet, a low-carb, low GI diet had no effect on
insulin sensitivity, systolic blood pressure, LDL-cholesterol,
or HDL-cholesterol but did lower triglycerides by 23%.
It is important to remember that this study was composed
of overweight and obese people with elevated blood pres-
sure (prehypertension or stage I hypertension), who were
consuming variations on a very healthy diet that included
a large number of fruits and vegetables. This study did not
address the effects of glycemic index on a typical western
diet, or in normal weight people.
While it is obvious that the GI values of foods will change
when combined into meals, this study’s diets produced the
expected differences in blood glucose AUC over 12 hours.
  Focusing dietary
recommendations
simply on the
glycemic index
appears to be an
oversimplification
of the multifactorial
process involved
in healthy blood
sugar control.
Both a lower GI value and a lower carb content were able
to lower the AUC compared to a high GI, high-carb diet.
Interestingly, there were no additional benefits on blood
glucose from lowering both the GI and the carb content.
This can be referred to as a plateau effect.
Insulin index is another part of this equation, which refers
to the insulin response a particular food generates. While
this is highly correlated with carbohydrate content, Figure 2
shows that some types of foods have either lower or higher
insulin responses than would be predicted from their glyce-
mic responses. Specifically, foods that are high in protein, as
well as refined bakery products, may cause a disproportion-
ately high insulin response.
Figure 2: Mean insulin index vs glycemic
index of different food types
Additional research has suggested that glycemic index val-
ues may be related as much to clearance of glucose as they
are to entry of glucose into the bloodstream. Focusing
dietary recommendations simply on the glycemic index
appears to be an oversimplification of the multifactorial
process involved in healthy blood sugar control.
Although LDL-cholesterol was higher in the high-carb, low
GI group, the researchers did not measure LDL-particle
63

  In the context of
a diet rich in fruit
and vegetables
and low in
processed foods,
glycemic index
may not matter a
whole lot.
size or LDL-particle number. It is possible that these chang-
es in LDL-cholesterol could be benign or even favorable.
Additionally, the low carb diets contained higher amounts
of protein than the high carb diets (23 vs 16% daily kcals),
and the high carb, low GI diet had the greatest amount of
fiber (37 g/day vs 29-33 g/day for the others). It is certainly
plausible that these factors could affect results, though one
might expect the effect of high fiber to be the opposite of
what was observed.
The big picture
“In the context of an overall DASH-type diet, using glycemic
index to select specific foods may not improve cardiovascu-
lar risk factors or insulin resistance.”
This quote from the study sums the findings up nicely.
However, the importance of the first part of that sentence
should not be underestimated. In the context of a diet rich
in fruit and vegetables and low in processed foods, glycemic
index may not matter a whole lot.
Furthermore, high GI foods may even be favorable to low
GI for certain outcomes when eating a DASH-type diet. It
remains to be seen whether or not using glycemic index to
select foods in the context of a typical western diet (and in a
well controlled study such as this) would improve CVD risk
factors.
Frequently asked questions
Protein and fiber levels were different between groups. Does
that change interpretation of the results?
It is possible, however the study was designed to make
comparisons between high and low GI, within the context
of either a high or low-carb diet. Because the protein was
the same within low-carb or high-carb groups, this should
not affect the outcomes. Fiber was highest in the high-carb,
low GI group, which was inevitable to some degree as fiber
content is a contributor to lower GI values.
How does this relate to “glycemic load”?
Glycemic load is glycemic index that has been adjusted to
reflect typical serving size. This is useful because we rarely
consume 50 grams of every food. Some foods (like grains,
for many people) are consumed in greater amounts, while
something like a carrot will have a high glycemic index but
low glycemic load. Here is a thorough resource for glycemic
load values.
How was the participant adherence to the protocols?
The researchers reported very high adherence, as all of
the food was provided to the participants and they were
instructed to write down any additional foods they may
have consumed and any of the food given to them that they
did not eat.
All food was consumed and no additional foods were eaten
on 96% of person-days on each diet. Although this was not
intended as a weight-loss study, participants lost about one
kilogram of body weight from baseline, with no differences
in weight loss between any of the groups.
64
I’ve [got a history of diabetes in my family] [am prediabet-
ic] [am diabetic]. Should I eat low glycemic? Should I try
the DASH diet?
There are arguments on both sides. On the one hand, the
DASH diet has been shown in meta-analysis to improve
certain markers of blood sugar control. While evidence is
mixed, long term observational data suggests that a lower
GI diet can reduce the risk of some chronic diseases, with
trials also showing a reduction in inflammatory markers.
However, the randomized trial under review suggests that
theorized benefits may be overblown. Observational data
 micromanaging
may not mean much, as people who eat lower GI could
have a variety of other characteristics that help them with every meal based
on glycemic index
disease prevention.

Also, micromanaging every meal based on glycemic index

of specific foods is likely to be more annoying than helpful.
Fat and fiber are typically present in meals consisting of
of specific foods is
minimally processed foods, and can blunt high blood sugar
responses. The glycemic load of a mixed meal can be hard
likely to be more
to estimate. If you are diabetic and on medication, listen
to your physician and other health professionals, and don’t
annoying than
rely solely on your own research. But following a named
diet isn’t a prerequisite for controlling your blood sugar. helpful.
What I should know?
In the context of an overall healthy diet that is rich in
fruits and vegetables, the glycemic index of foods is not
something to be very concerned about. This results of this
study even suggest that in the context of a higher-carb diet,
high-GI foods may be preferable to low GI foods from the
standpoint of their effects on insulin sensitivity and LDL-
cholesterol levels. ◆

Got some thoughts on this study, or the glycemic index in

general? DASH over to the ERD Facebook group and let us
know what you think.

65
 INTERVIEW:
Ivan Oransky
Ivan Oransky, MD, is the vice president and global editorial director of MedPage Today,
co-founder of the MacArthur Foundation-funded Retraction Watch, and founder of
Embargo Watch. He previously was executive editor of Reuters Health and held edito-
rial positions at Scientific American and The Scientist. A 2012 TEDMED speaker, he was
awarded the 2015 John P. McGovern Award for excellence in biomedical communica-
tion from the Southwest Chapter of the American Medical Writers Association. He has
authored or co-authored four books and written for numerous publications, including
Nature, The New Republic, and The New York Times.

I believe Retraction Watch is turning five later on this year. Considering the massive number of retractions
you've reported on, what are a couple of your favorite ones? Like ones that might make for a good made-for-
TV movie (with "good" loosely defined).
Retraction Watch does indeed turn five in August. Good memory! My favorite stories often change over
time, but right now those that involve fake peer review – often reviewed by the very authors who wrote the
papers – are still at the top of my list. It might be difficult to turn those stories into a made-for-TV mov-
ie, but we’ve turned them into a longer feature for Nature. They’re a perfect mix of pressure to publish and
ingenuity, in the same way someone might construct a perfect crime – but then get caught. For a more
typical made-for-TV movie script, though, I’d go with the story of Karel Bezouska, who broke into a lab one
night to tamper with an investigation into his work, or Jatinder Ahluwalia, who sabotaged his colleagues
and later turned out to have been dismissed from an earlier PhD program for the same reasons he eventual-
ly had to leave a second university.

Why do you think the number of retractions has been going up over time?
You’d expect to see more retractions as there are more papers published, but as Nature pointed out in 2011,
the rise in the rate of retractions from 2001 to 2010 far outstripped the growth in papers. Part of that was
definitely due to the advent of plagiarism detection software, and the fact that there are more eyeballs on

66
papers now that journals are all online. There’s also at least
circumstantial evidence that the amount of misconduct is
going up, too. That’s less clear, though.
Examine.com was plagiarized last year, and the authors
became extremely defensive. Have you interacted much
with authors of retracted articles? What are these interac-
tions like?
We always try to contact the authors of retracted papers.
Some get back to us. When they’re retracting for hon-
est error, those conversations show the best of science’s
self-correcting nature. We often add them to our “doing the
right thing” category. Among those authors retracting for
misconduct, or what sounds like misconduct, some don’t
respond to our requests, while others don’t have much to
add, or give us a few details that flesh out the story. Most
of the time with those kinds of retractions, we have to rely
on other sources for information. And then there are the
handful of authors who try to keep themselves out of the
limelight by threatening to sue us. It’s never worked – quite
the opposite. Just Google “Streisand Effect.”
  Karel Bezouska, who broke into a lab
one night to tamper with an investigation
into his work, or Jatinder Ahluwalia,
who sabotaged his colleagues and later
turned out to have been dismissed from
an earlier Ph.D. program for the same
reasons he eventually had to leave a
second university.
I'm not sure how often meta-analyses account for included
studies that have been retracted. Do you have any sense
of how that issue is dealt with? After all, there isn't a duty
for meta-analysis authors to follow-up on included studies,
especially if they are numerous.
That’s a critical issue, but I’m not aware of anyone looking
at it systematically, so to speak. It came up in meta-analyses
performed after anesthesiology researcher Joachim Boldt’s
fakery came to light. Most Cochrane meta-analyses are
regularly updated, so perhaps there should be a mechanism
that alerts meta-analysis authors to retractions and helps
them figure out whether they need to follow up. In fact, we
were just awarded a $400,000 grant from the MacArthur
Foundation to build a retraction database that we hope will
be linked to personal libraries authors use for referenc-
es. That would make it impossible to cite a retracted paper
without first knowing that it had been retracted.
If you were magically given authority over retraction poli-
cies for all journals, what changes would you implement?
This seems unlikely, but since you’ve invoked magic, I’ll
67
make pretend: First, I’d insist on clear notices that detailed
why exactly a paper had been retracted. No more opaque
“This paper has been withdrawn by the authors.” Of course,
that would probably require hiring different lawyers than
the ones many publishers seem to employ today when they
cave to legal threats from fraudsters. I’d also do away with
the mulligan of retractions, the unexplained “withdrawal”
of papers that are published online but not in print yet. The
idea that such a paper doesn’t exist enough to warrant a
full retraction notice is a quaint but ridiculously outdated
notion some publishers insist on.
Do high-impact journals deal with retraction in a more
transparent manner? Are there any journals that do an
exemplary job of it?
High-impact journals generally provide more informa-
tion than others in retraction notices, but they also have
more retractions per thousand papers published. I’d say the
Rockefeller University Press journals such as the Journal of
Cellular Biology do a consistently above-average job. Keep
in mind, though, that how well a given journal investigates
allegations is usually a black box, unless we happen to learn
of it from a source, and we’ve seen lots of journals stubborn-
ly refuse to correct the record, so it’s hard not to think that
the notices we see are just a tip of the iceberg phenomenon.
Back to the human angle: I can't imagine being the author
of a particularly egregious retracted article, and then
attempting to procure more grant money and get more arti-
cles published. For academics who live and die by their CV,
how can they continue in academia?
That’s an excellent question, and we have to say we’re always
surprised by a vocal minority of our readers who object
strenuously when we ask it by following up on cases of
fraud. In fact, more than half of researchers found to have
committed misconduct by the Office of Research Integrity
are still in the positions they had when they did the deed.
That strikes us as strange, given how competitive grants are.
It is a reminder, however, of how reluctant many researchers
are to blow the whistle or air science’s dirty laundry. They
may think that’s a good way to minimize the effects of fraud,
but in the long run it’s going to erode trust. Better to be
forthright about cases of fraud, which are still uncommon,
then pretend they don’t exist and then have to defend sci-
ence when news breaks.
We've lightly touched on journals and information dis-
semination in previous issues, talking about predatory
publishing and exaggeration in press releases. What's on
your mind these days concerning information dissemina-
tion from journal articles? For example, I've read some of
your writings on embargoes and the Ingelfinger Rule.
I’ve more or less said my piece on how to use embargoes, if
journals must, but I remain concerned that the Ingelfinger
Rule prevents journalists from telling the real stories of
science. I’ve said more about this elsewhere, but basical-
ly it keeps researchers from talking to reporters for fear
of having their work rejected by top journals. You’d think
the rough-and-tumble world of online journalism would
have eroded some of Ingelfinger’s power, but if anything
it’s strengthened it as reporters look for the next minute’s
“scoop.” Still, I’ve seen some chinks in the Ingelfinger armor,
and that makes me happy. ◆
Dr. Oransky earned his BA at Harvard and his MD
at the New York University School of Medicine.
While a student, he served as executive editor of
The Harvard Crimson and as co-editor-in-chief
of the medical student section of the Journal
of American Medical Association. In addition
to his work as a medical writer and editor, Dr.
Oransky teaches medical journalism at New York
University’s Science, Health, and Environmental
Reporting Program, and he is a clinical assistant
professor of medicine at New York University
School of Medicine. Dr. Oransky currently serves on
the board of directors and as vice president trea-
surer of the Association of Health Care Journalists.
68
INTERVIEW:
Jessica Richman
Jessica Richman started and sold her first company after high school. At Stanford
University she earned degrees in Economics and Science, Technology and Society
(with a computer science focus). Along the way, she worked for Google, McKinsey,
Lehman Brothers, the Grameen Bank, and top-tier Silicon Valley venture firms.
Jessica arrived at Oxford University as a Clarendon Scholar and completed an MSc at
the Oxford Internet Institute. She is currently a DPhil student at Oxford with a focus
on innovation, social networks, and collective intelligence.

In layman's terms, how does uBiome go about testing for bacteria? What can you detect and what
can't you detect? And also...is the testing technology changing from year to year?
Great question, and a warm hello to all inquiring Examine.com readers. uBiome offers kits for peo-
ple to sample the bacterial populations anywhere on the body and do personal experiments to learn
about themselves.

Sampling is done from the comfort and privacy of your own home, with a simple and non-intrusive
procedure. For the gut sample, all we need is a swab of your toilet paper; for the skin sample, it’s just a
swab behind your ear. We also sell kits for nose, genitals, and mouth.

Each site has a unique microbiome that is home to a specific balance of flora. With the samples from
each of these body sites we can tell you about the bacteria that inhabit that microbiome. You also get
access to our ever-expanding database of bacteria, to see the role those bacteria play in keeping your
microbiome balanced.

You can track the results of probiotic usage, diet and supplement experiments, and antibiotic treatments,
so you can see how they affect your microbiome over time. You can also compare your results to other
participants, to see how you compare to users who have made similar dietary or lifestyle choices.

69
This is an exciting, emerging field. Our team keeps up to
date on the many studies that are published about the
microbiome and constantly works on ways to make this
data accessible to our participants - regardless of their sci-
entific background. We want to help people engage with
their data in the most meaningful way possible.
We've covered a couple rodent studies looking at microbi-
ome effects on health and performance. Conclusions from
human research are a bit more limited, due to complexity
and inability to control as many
parameters. Where does uBiome
fit into the research picture?
uBiome is the largest citizen
  What you
science research study looking
at the microbiome. We aim to
eat impacts not
contribute significant findings
to this relatively nascent frontier only your gut
flora but your
of scientific research. While our
test is not diagnostic, we do want
to empower our community of
knowledge-seekers with detailed
information and personalized
microbiome at
visualizations that will help them
understand how their microbial
other sites too.
balance relates to other data sets.
One of the most exciting ways uBiome fits into the micro-
biome research picture is by equipping people with tools
to create their own scientific studies so they can explore
the issues most important to them. We want to see people
engaging with their microbiome in ways that might not
have been possible before. We want to learn what they learn.
The microbiome is much more complex than most people
can fathom. Some bacteria can be "bad" in one context
and not so bad in another. So recommendations are likely
to be more complicated than for things like vitamins and
minerals. What advice do you give to gut novices about gut
health? What about people who are already somewhat edu-
cated on the subject?
For people who are new to this area, we want to help them
understand that the microbiome is influenced by both
health and lifestyle choices. What you eat impacts not only
your gut flora but your microbiome at other sites too.
More than 1.4 billion adults worldwide are overweight, with
more than 500 million clinically obese. Over the past two
decades, there has been a marked rise in allergies, autoim-
mune, and neurological disorders. Autoimmune disorders
now affect 23 million Americans alone; Asthma in western-
ized societies has risen steadily,
doubling in the last 20 years.
There is overwhelming scien-
tific evidence that the human
microbiome has profound con-
sequences for our health, and is
implicated in diseases ranging
from obesity, diabetes, and IBD
to heart health, anxiety, and
depression.
uBiome is designed to help
explore the scientific ramifica-
tions of the microbiome and to
better understand this landscape
for the novice and expert alike,
but we can’t give people specific recommendations on what
to eat based on your microbiome.
What can people currently learn by having their various
microbiomes sampled? What might they be able to learn
five years down the road (especially with continued testing)?
With uBiome sampling, people can learn:
• What proportions of different bacteria are in their
microbiome
• How they compare to the microbiomes of other rel-
evant groups, such as healthy omnivores, vegans,
smokers, and people who exercise regularly.
• How their data compares to peer-reviewed studies of
the microbiome
70
   Papers show that antibiotics or a
drastic change in diet can alter a person’s
microbiome in as little as three days.
Testing provides participants with a snapshot of their micro-
biome at a single point in time. However, we so want to
keep our community informed about the studies that scien-
tists are only now beginning to develop. Sampling now and
at regular intervals will allow people to have early access to
emerging studies, and give them the means to understand
how lifestyle or dietary choices may have changed their
microbiome over time.
How quickly can people change their gut microbiomes? For
example, after a couple courses of antibiotics. Have you
seen any particularly interesting success stories?
Studies are just beginning to explore how susceptible the
microbiome is to change. Papers show that antibiotics or a
drastic change in diet can alter a person’s microbiome in as
little as three days.
Our participants are not just simply learning about their
bacterial balance - many are setting out to explore the
extent to which they can hack their own microbiome. We
are doing some internal studies on this -- stay tuned! And
definitely let us know if any of you would like to join in, by
writing to us at studies@ubiome.com.
Most people focus on the gut, but uBiome samples five sites
on the body. What are some things people should know
about these other microbiomes?
The important thing to realize is that your microbiome
functions differently at each of these sites.
For example, scientists who study the oral microbiome
are currently examining the role of bacteria such as
Streptococcus mutans in tooth decay. A healthy mouth
microbiome can help fight off the bacteria that cause cavi-
ties or gum disease.
The vaginal microbiome is also a rich area of research, with
new insights being generated frequently.
The FDA cracked down on 23andme last year. What sort of
regulatory issues does uBiome face, if any?
uBiome is a research study, and we do not give any diagnos-
tic information to consumers.
One of the most frequently asked questions we receive is
if we are IRB approved. For anyone not familiar with the
term, an Institutional Review Board is a committee that
approves, oversees and reviews biomedical and behavioural
research involving humans. They conduct analysis to deter-
mine whether or not research should be done. Their priority
is to advocate for the individual participant.
To answer that question here: uBiome has received research
study approval from E&I Review Services. To be complete-
ly sure that we had our participants’ privacy, security and
well-being in mind, we have in fact consulted with our IRB
every step of the way. Under their advice, we completed the
final stages of their analysis and received approval immedi-
ately after we had reached our initial funding goals.
Readers of ERD are often scientists or wannabe scientists.
Lots of information junkies. With so many different sourc-
71
es of information, and so many things that can be tracked,
what are some of the most interesting developing microbi-
ome topics that we should keep in mind? Any must-read
gut-centered websites or blogs?
  [...] question
There is a wealth of information available for people who
want to learn more about the human microbiome. With
to your readers:
websites such as Google Scholar becoming increasing-
ly popular, it is becoming easier for people to access that
what are the
information all the time.
most burning
Institutions like MIT also offer excellent courses online
for free. People who are interested in learning more about microbiome-
related questions
human biology, before they dive into their microbiome,
might be interested to take their courses here.

The National Institutes of Health is a great source for

gut-centered material. For a lighter read, check out
on your mind, or
MindBodyGreen.com. And Paleo Magazine has some
interesting articles and podcasts.
what’s the number
We’re also firing up the uBiome blog, and would love to end
one thing you’d
with a question to your readers: what are the most burning
microbiome-related questions on your mind, or what’s the want to learn from
looking at your
number one thing you’d want to learn from looking at your
microbiome?

Thanks so much for this opportunity to share our project

with the ERD community. ◆
microbiome?

Jessica is a co-founder of uBiome, the world's largest successful citizen science

project. UBiome has been featured in Wired, MIT Technology Review, Scientific
American, NPR, FoxNews, ABC News, and dozens of other media outlets. Jessica is
also founder of Sciencecitizen.org, an organization that helps researchers incorpo-
rate the public into their work.

72
 ERD
Until Next Issue...

The next issue of ERD will come out the

first week of March. In the meantime:

Comments? Questions? Join the

Concerns? Discussion Online
Send your thoughts to Join the conversation on
erdeditor@examine.com Issue 4 in our exclusive
ERD Private Forum
on Facebook

Spread the Word

73
Credits
Copy Editor: Dmitri Barvinok

©iStock.com/ilustro
©iStock.com/RKaulitzki
©iStock.com/Andrii_Oliinyk
©iStock.com/Dacian_G
©iStock.com/MackoFlower
©iStock.com/barol16
©iStock.com/Mari
©iStock.com/duckycards
©iStock.com/jophil
©iStock.com/DebbiSmirnoff
©iStock.com/Lisovskaya
©iStock.com/masaltof
©iStock.com/wsfurlan
©iStock.com/Nikada
©iStock.com/mel-nik
©iStock.com/Dole08
©iStock.com/carlosdelacalle
©iStock.com/lleerogers
©iStock.com/lzf
©iStock.com/lvenks
©iStock.com/denisgo
©iStock.com/dtimiraos

74