Thursday, June 30, 2016

A Letter to the Editor Criticising the Paleo Mouse Study

A paper was published in February 2016 claiming that they tested the effect of a Paleo diet in mice and found that it causes excess weight gain, impaired glucose tolerance and insulin resistance [1].  I dubbed the paper ‘the Paleo mouse study’ and have written about it earlier in the year.  I discussed that mice respond differently to higher fat diets than humans*.  And also, that the LCHF diet is the Paleo mouse study sucks as it was largely comprised of added fats, casein and sucrose whereas the standard low fat diet was a lot more whole foods based

A letter to the letter by Nathan Cofnas** was recently published that brought up some issues with the Paleo mouse study [2] (it’s short, open access and I recommend reading it):

  • The representation of the study in the media*** and on Melbourne University’s Youtube channel was that this study was a test of a LCHF Paleo diet (never mind the details on what the mice actually ate), but the paper didn’t even include the word ‘Paleo’ or ‘Paleolithic’
  • The Paleo concept is based on evolution and genetic adaptation and would make the hypothesis that animals do best on the diets that they are most genetically adapted to, which can be largely inferred from what they eat in the wild.  For mice, this is a low fat, high carbohydrate, largely plant based diet, so it would be expected that the chow diet would be better for them 

But one thing the letter didn’t mention much of are results from human RCTs, which trump and contradict and mouse studies.  Meta-analyses of low carb vs. low fat RCTs [3] or Paleo vs. conventional dietary advice RCTs [4] both find that low carb and Paleo diets result in greater weight loss than the alternative, and more recent studies continue to support these findings [5, 6]

* For example, the senior author of the Paleo mouse study recommends the Mediterranean diet, but the amount (~40% of total calories) and type (olive oil/MUFA) of fat promoted in the Mediterranean diet causes obesity, glucose intolerance and insulin resistance in mice [7, 8]

** He is part of the Department of History and Philosophy of Science at the University of Cambridge and has published some earlier work on evolutionary mismatch [9].  Leave it academics outside the field and bloggers to apply some common sense in translating studies to the real world 

*** For example: “To put that in perspective, a 100 kilogram person on a Paleo diet could pile on 15 kilograms in two months” [10] (good luck achieving half that effect with a deliberate overfeeding study in humans)

Wednesday, June 29, 2016

Moderation is Subjective and Susceptible to Bias

I discussed the concept of moderation in an earlier post.  My problems with moderation as dietary advice include: that moderation is poorly defined and very subjective; it can be used by the food industry to legitimise their unhealthy products; and some people do better with abstinence than with moderation

The problems with the first point – the subjectiveness of moderation – was well illustrated in a recent paper [1].  In the introduction, this paper cites earlier research that found: (1) people are poor judges of food intake; (2) people look to their peers and themselves to determine what ‘moderate’ consumption should be; (3) people’s beliefs are biased to favour themselves

Study 1

The first study involved asking female students about how many cookies: (1) one should eat; (2) would be moderate consumption; and (3) would be considered indulgent

On average, the participants reported that they should eat 2.25 cookies and that a ‘moderate consumption’ was 3.17 cookies

Study 2

The second study involved asking people taking a survey how many candies would be: (1) a reasonable amount to eat in one sitting; (2) eating in moderation in one sitting; or (3) what they should eat in one sitting.  Then they reported how much they like the candies

On average, the participants reported they should eat 8.87 candies, moderation consumption was 10.93 candies, and a reasonable amount was 14.17 candies.  There was also a correlation between the number of candies reported as moderate consumption with the liking of the candies (r = 0.38, p < 0.0001) and the reported consumption of those candies (r = 0.27, p = 0.007), whereas correlations were weaker or not significant for ‘should eat’ and ‘reasonable amount’

Study 3

The third study involved asking people taking a survey to report their average consumption of various foods and how moderate they thought their consumption of those foods

On average, the participants rated their consumption as moderate (4.50 and 4.47 on a 1-7 point scale, where 1 = ‘not at all’, and 7 = ‘very much’).  Across the 12 food and beverage categories, participants on average defined moderate consumption as greater than their personal consumption

There was also a correlation between reported personal consumption and what was reported as ‘moderate consumption’ (r = 0.52, p < 0.001 and r = 0.50, p < 0.001).  This correlation wasn’t disproportionate seen in people who were overweight.  However, there was no correlation between (1) the participants’ personal consumption with the participants’ rating of their consumption as moderate; and (2) the participants’ rating of their consumption as moderate with how much they believed was moderate consumption


Moderation could work as a concept, but not in situations where people’s reference point is one where they should be eating unhealthy food and that moderation involves eating a greater consumption of unhealthy food; or where most people believe they are already doing so well (when most people clearly aren’t) that moderation would once again involve a greater consumption of unhealthy food

For moderation to work, I think you would need to define ‘should’ as 0 and moderation as only slightly above 0, and for most people for moderation to mean something closer to self-discipline than a self-rationalised or socially acceptable indulgence 

If you want to adopt the recommendation of moderation, then I suggest you acknowledge then try to account for your biases and operationalise what moderation means.  Then most importantly, make adjustments based on the feedback your body is giving you

Tuesday, June 28, 2016

Other Publications in Peer-Reviewed Journals

After the publication of my honours project, I think this is a good time to summarise some other papers I’ve contributed to (as a general rule the first author made the most contribution and the last is the senior author).  These papers aren’t open access, but you can find them all at ResearchGate (links included).  You can also find them on PubMed by searching ‘hamley s’ (the perks of having an uncommon last name)

Overexpression of sphingosine kinase 1 in liver reduces triglyceride content in mice fed a low but not high-fat diet [1]


  • Half the mice were put on a high fat diet (HFD) and half on a standard low fat chow diet (LFD)
  • Some of mice were injected with a non-infectious virus to increase the expression of sphingosine kinase 1 


  • The HFD group developed obesity, impaired glucose tolerance, insulin resistance and fat accumulation in the liver
  • The HFD group had lower expression of sphingosine kinase 1 but not sphingosine kinase 2
  • Overexpression of sphingosine kinase 1 reduced liver triglycerides, de novo lipogenesis and lipogenic gene expression in the LFD group but not in the HFD group
  • Overexpression of sphingosine kinase 1 did not alter glucose tolerance or insulin sensitivity in mice on the LFD or HFD groups 

Implications: The idea of this study was that overexpression of sphingosine kinase 1 may reduce lipid accumulation in the liver, which would protect against the development the development of glucose intolerance and insulin resistance in the HFD group.  The first step on this process didn’t happen in the HFD group, which was unexpected.  So the fact that glucose intolerance and insulin resistance wasn’t altered in the overexpressing HFD group shouldn’t be surprising

Application of dynamic metabolomics to examine in vivo skeletal muscle glucose metabolism in the chronically high-fat fed mouse [2]


  • Half the mice were put on a HFD and half on a LFD
  • An OGTT was used with the glucose coming from a glucose isotope with all carbon atoms having an extra neutron (U-13C).  The use of U-13C glucose enables the metabolic fate of the ingested glucose to be tracked 


  • The HFD group developed obesity, impaired glucose tolerance and insulin resistance
  • The HFD group had less heavy carbon labelling in 3PGA (an intermediate of glycolysis), lactate, alanine and the TCA cycle, except for citrate 

Implications: The labelling data suggests that glucose metabolism in muscle is impaired with insulin resistance.  The labelling data also indicates that almost all the pyruvate enters the TCA via pyruvate dehydrogenase (PDH, cataplerosis) rather than pyruvate carboxylase (CK, anaplerosis)

In vivo cardiac glucose metabolism in the high-fat fed mouse: Comparison of euglycemic-hyperinsulinemic clamp derived measures of glucose uptake with a dynamic metabolomic flux profiling approach [3]


  • Half the mice were put on a HFD and half on a LFD
  • An OGTT was used with the glucose coming from a glucose isotope with all carbon atoms having an extra neutron (U-13C)
  • The mice were infused with insulin and glucose to maintain elevated insulin levels and fasting glucose levels (hyperinsulinemic-euglycemic clamp) 


  • The HFD group developed obesity, impaired glucose tolerance and insulin resistance
  • Insulin resistance in the heart was present at 3 weeks after being on the HFD and does not get worse over time.  This is similar to muscle, but the liver develops insulin resistance earlier (present at 1 week)
  • There were no significant differences between the diets in heavy carbon labelling in the intermediates and products of glycolysis and TCA cycle, except for lower alanine in the HFD group 

Implications: The labelling data suggests that glucose metabolism in the heart isn’t reduced.  This is likely to be due to the higher glucose levels, which compensates for the insulin resistance.  Why this happens in heart but not muscle is unknown.  The labelling data also indicates that most of the pyruvate enters the TCA via PDH rather than PC anaplerosis.  The heart is a relatively small contributor to energy expenditure and whole body insulin sensitivity/glucose uptake, but this might be important for some heart diseases

Sunday, June 19, 2016

My Honours Project: Pre-Diabetes and Insulin Resistance Completely Normalised after Just 1 Week

My honours project recently got published in a peer-reviewed journal [1].  Academic publishing can take a while, but it gets done eventually.  It’s a good project and may be of interest, but at times it’s not quite written for the general public, so I’ll summarise it here:


We put 20 mice on the standard laboratory low fat chow diet (CHOW) and put 40 mice on a high fat diet (HFD)*.  After 8 weeks we switched 20 mice on the high fat diet to the chow diet (HFD→CHOW).  We did oral glucose tolerance tests (OGTT) before and after the diet switch, and took tissues at the end of the study

We used:

·         OGTTs to measure glucose tolerance and how insulin and free fatty acids changed during the OGTT
·         Stable isotopes of glucose to measure the source of plasma glucose during the OGTT – whether it came from the body (normal glucose) or from the OGTT (heavy glucose).  This can give an indication of whether the defect is from excess glucose production or impaired glucose uptake by tissues
·         Heavy water (2H2O) to measure lipid synthesis and the sources (but not the absolute amount) of glucose production (glycogenolysis vs. gluconeogenesis)
·         Tissue samples to measure glycogen, triglycerides, other tissue lipids, lipid synthesis and various metabolites (metabolomics)

* I’ve written before about how the mice strain we used (C57Bl/6) was bred to be susceptible to develop obesity and the metabolic syndrome on a high fat diet [2].  So think of this as a model of diet induced obesity and pre-diabetes rather than being prescriptive for human diets


When the mice were on the high fat diet:

·         They ate more calories (but slightly less volume, data not shown) (figure 1)
·         They gained more weight as fat mass (and had no change in the weight of their quads) (figure 1)
·         They developed impaired fasting glucose and impaired glucose tolerance (pre-diabetes) and insulin resistance (figure 2)
·         They developed fat accumulation in their liver (figure 5) and muscle (data not shown)

When the mice were switched back to standard laboratory chow they

·         Substantially reduced their energy intake voluntarily and even ate less than the chow group, and as a result lost about half their excess weight in 9 days (figure 1)
·         After 1 week they completely normalised their pre-diabetes and insulin resistance (figure 2)
·         Lost some, but not all the fat in their adipose tissue (figure 1), liver (figure 5) and muscle (data not shown)
·         Got more accumulation of some ceramide species in the liver (ceramide accumulation is a proposed mechanism of insulin resistance, so this surprised me) (figure 5)

The Energy State as a Mechanism of Pre-Diabetes and Insulin Resistance

A potential mechanism of pre-diabetes and insulin resistance could be that they are caused by an elevated energy state in the liver (low AMP or a low AMP:ATP ratio) as a consequence of excess energy intake:

Excess energy intake >>> elevated hepatic energy state >>> inhibited glucose metabolism and insulin resistance >>> hyperglycemia

This is based on:

·         Changes in fasting glucose, glucose tolerance and/or insulin resistance occur within ≤ 1 week in humans and mice in response to overfeeding, calorie restriction and bariatric surgery.  So you need a mechanism that can be manipulated very quickly and the hepatic energy state can be
·         The energy state of the cell negatively regulates glucose metabolism [3]
·         Previous research that found the hepatic energy state was elevated in HFD mice

We used targeted metabolomics to measure AMP and the intermediates of glycolysis (G6P, F6P, 3PGA, PEP) in the liver.  The HFD group had lower AMP and intermediates of glycolysis, but higher glucose in the liver* (figure 4).  The higher glucose and lower concentrations of glycolytic intermediates could suggest impaired glucose metabolism, but this isn't a imprecise measure as differences in static concentrations don’t necessarily parallel differences in metabolic fluxes

However, the HFD→CHOW group only had a partial normalisation of AMP and intermediates of glycolysis (figure 4).  Considering they completely normalised their pre-diabetes and insulin resistance, this casts doubt onto the mechanism

* The primary glucose transporter is in the liver GLUT2, which doesn’t require insulin and has a high capacity and low affinity for glucose.  This means the glucose concentration in the liver will be similar to the blood, so this result is to be expected, and this enables the liver to act as a glucose sensor (like the beta cells of the pancreas)


Metabolic stuff generally happens very quickly.  This is the case whether you’re talking about cholesterol [4], blood pressure [5], or in this case, pre-diabetes and insulin resistance (which I’ll discuss more of later).  The rapid timeframe and results of this study suggest that mechanisms relying on overweight/obesity and fat accumulation in either the liver or muscle are probably not major mechanisms in the development and normalisation of pre-diabetes and insulin resistance

Whatever the mechanisms are, they have to be capable of acting very quickly.  Based on some plausibility and the timeframe, some possible mechanisms could be based in the brain, liver and gut, but probably not muscle.  More on this later