Wednesday, December 31, 2014

Paleolithic Diet Trials: Boers, et al (2014)

34 people with at least 2 aspects of the metabolic syndrome were randomised to follow a Paleolithic diet (PD) or a diet based on the Dutch Health Council guidelines (DHCG) for 2 weeks.
At baseline there were some significant differences between the groups: the PD group weighed more (98.0kg vs. 86.0kg), had more aspects of the metabolic syndrome on average and more people with low HDL-C and high fasting glucose.  HDL-C, triglycerides, the triglyceride:HDL-C ratio, the total:HDL-C ratio and hs-CRP were almost significantly higher in the Paleolithic group (p = 0.06-0.11) (table 2 and table 3)
The Paleolithic diet “was based on lean meat, fish, fruit, leafy and cruciferous vegetables, root vegetables, eggs and nuts. Dairy products, cereal grains, legumes, refined fats, extra salt and sugar were not part of it”.
The participants were given daily menus for the trial period, which were calorie controlled to 2078 calories (so the diets were isocaloric).  “Although coffee and tea were not part of a Palaeolithic-type diet, subjects were allowed to drink, in view of possible withdrawal symptoms, up to two cups of coffee or black tea per day. Any medication was continued at the same dosages”
Nutrient intake is presented in table 1 and I summarised the calorie and macronutrient ratios in the table below.  Micronutrient intake was also measured and the PD group had a high intake in 15/19 of micronutrients measured (including LCO3, excluding sodium and vitamin D), except phosphorus (equal), calcium, iodine and vitamin B3 (lower), see graphs below
Although we weren’t told what both groups actually ate, we can have a good guess.  Since the PD group had a higher intake of animal protein and most of their carbohydrates came from sugars they were probably consuming more meat and fruit, with starchy vegetables being a minor part of the diet
Energy (kcal)
Nutrient intake (% of RDA)
Average: PD 173%, DHCG 117%
Median: PD 143%, DHCG 118%

 One of the aims of the study was to keep body weight stable.  However, 9 participants (7 in PD group, 2 in DHCG group) had over 2kg of weight loss and so were given extra snacks even though they weren’t hungry.  This could indicate a change in the body fat setpoint (leptin resistance)
Both groups had a number of improvements, specifically in waist circumference, blood pressure, fasting glucose and blood lipids (table 3, table 4 and table 5).  Neither group had a significant change in AUC glucose, AUC insulin, hsCRP, TNFα and the lactulose:mannitol ratio (measure of intestinal permeability).  Also see the table below
Dutch Health Council Guidelines
Abdominal circumference (cm)
Systolic BP (mmHg)
Diastolic BP (mmHg)
Glucose fasting (mmol/l)
Insulin fasting (mU/l)
AUC insulin (mU/l x min)
TG:HDL-C (mol/mol)
TC (mmol/l)
HDL-C (mmol/l)
LDL-C (mmol/l)
TG (mmol/l)
TC:HDL-C (mol/mol)
Homocysteine (μmol/l)*
Bold: p < 0.05, underline p < 0.10
After adjusting for weight loss, the improvements in systolic blood pressure, characteristics of the metabolic syndrome, TG:HDL-C, HDL-C, TG and TC:HDL-C were significantly greater in the PD group relative to the DHCG group (table 6)
Overall, this trial found a Paleolithic diet is more successful than convention dietary guidelines at improving features of the metabolic syndrome even though the diets were isocaloric.  The main limitations though are the small number of participants (n=34) and the higher starting weight and worse health of the PD group at baseline gives them more room for improvement.  The short duration of the trial mainly makes it difficult to access long term compliance.  It’s worth noting that: “after two weeks 89% of the Palaeolithic group and 64% of the reference group were still motivated to continue their dietary regimes”.  I don’t see the short duration of the trial being a limitation in so far as getting improvements in metabolic health as things like glucose tolerance and insulin resistance can be altered (even completely restored) in much shorter timeframes.
* It’s odd that homocysteine significantly increased in the PD group, but not the DHCG group even though folate and vitamin B12 intake were higher in the PD group

Monday, December 29, 2014

Does Saturated Fat Impair Endothelial Function?

Saturated fat (SFA) is thought to impair endothelial function.  As far as I can tell there are two clinical trials looking at how endothelial function is affected when SFA is replaced with other macronutrients:
Study 1:
This study used a crossover design, where 40 healthy people were randomly allocated to either a high SFA, MUFA, PUFA or a low fat, high glycemic load (GL) diet for 3 weeks each diet.  The participants were given ‘test’ foods to manipulate the macronutrient composition of the diet.  The macronutrient composition of these test foods can be seen in table 1.
“Subjects returned at the end of each intervention for a fasting venous blood sample and weight, FMD, and PWV measurements.”
Test Food
50g Butter
20g High MUFA Margarine
45g Almonds
20g High PUFA Margarine
35g Walnuts
70g Sultanas
50g Jam
The high SFA diet reduced flow mediated dilation (FMD) (measure of nitric oxide bioavailability/endothelial function) (figure 1), didn’t alter pulse wave velocity (measure of arterial stiffness) (table 4) and increased some markers of inflammation (table 4 and figure 3).
They also measured blood lipids and insulin (table 3).  The effects on blood lipids were predictable and there were no significant differences in fasting insulin.  They didn’t find an association between blood lipids and FMD
Study 2:
This study put 121 people with mild insulin resistance on a high SFA diet (>15%) for 1 month, then randomised them to a high SFA, MUFA or carbohydrate diet for 24 weeks.  The difference between the diets is described in the methods section, which I’ve summarised below.  The macronutrient composition of the diet can be seen in table 2
“Participants were advised before each visit for the measurement of vascular function to avoid strenuous physical activity, foods high in fat, caffeine, or alcohol on the previous day; subjects were provided with a list of foods to avoid and a low-fat evening meal (<10 g fat; 2–3 MJ) before fasting overnight from 2200.”
“After the completion of an intravenous glucose tolerance test, as reported elsewhere (17), participants were provided with a low-fat meal (2 MJ; <5 g fat) and a drink of water, and measures of vascular function a ≥2 h later were made at St Thomas’ Hospital between 1400 and 1700; both baseline and follow-up measures were made at the same time of day to minimize diurnal variations.”
Test Food
Full fat milk and cheese
High SFA spread
Standard salad dressing
Skim milk and half fat cheese
High MUFA spread
High MUFA salad dressing
Nuts and potato crisps
Skim milk and half fat cheese
Reduced fat spread
Reduced fat salad dressing
Bread, potatoes and rice
There were no significant differences between the groups in all the measurements taken, which related to blood pressure, FMD, arterial stiffness and lipid peroxidation (table 3 and table 4).
* This trial also measured glucose tolerance and insulin resistance (by an intravenous glucose tolerance test rather than an oral glucose tolerance test), blood lipids and inflammatory markers, which was reported in another paper.  Once again, the effects on blood lipids were predictable and there were no differences on the other measures except that the low fat group lost weight and the change in weight was significantly different to the high SFA group, and the high SFA group significantly lowered their microalbumin:creatinine ratio (suggests improved kidney function) and the change was significantly different to the low fat group (table 2 and table 5)
The first study says SFA impairs endothelial function and second one says it doesn’t.  So what to do make of this?
In favour of the first study:
The first study’s methodology was more what I was expecting.  Whereas in the second study all the participants reduced their fat intake the day before and they did an intravenous glucose tolerance test and a low fat meal prior to the measurements of vascular function.  Both of these may have affected the outcome, particularly by reducing the differences between the groups.  It’s certainly odd that there were so few differences between the groups.
In favour of the second study:
Trying to isolate the effects of one nutrient in diet studies is very difficult.  The second study did this much better than the first study, which I thought did this quite poorly.  The first study can be viewed not as a comparison of macronutrients, but rather as a trial comparing butter to nuts and processed fruit.  This gives the MUFA, PUFA and GL groups in this study an unfair advantage as nuts and sultanas are whole foods and contain more micronutrients and other beneficial compounds than butter.  Even though the second study “…set out to test the hypothesis that decreasing SFA intake would improve vascular function. Our hypothesis was based on the belief that decreasing SFA intake would improve insulin sensitivity, which the main report showed not to be the case.”  They were critical of the first study saying that: “However, the interpretation of the results of that study were confounded by the use of high intakes of walnuts, almonds, and sultanas, which contain polyphenolic compounds that have antioxidant and other pharmacologic properties and can influence endothelial function (23).”  In support of this, adding walnuts to the diet improves endothelial function [2].
Does SFA impair endothelial function?  I don’t know, it’s hard to say from only two studies, particularly when each has problems.  I would guess probably not based on these studies and that there’s no good evidence that replacing SFA with MUFA, carbohydrate [3] [4] or PUFA [5] reduces cardiovascular disease.

Sunday, December 28, 2014

Macronutrient Myths from Alternatives

Excess Protein is converted to Glucose (Gluconeogenesis)
A widespread idea in the low carb community is that excess protein is converted to glucose by gluconeogenesis.  Therefore, many low carb advocates also recommend reducing/moderating protein intake to reduce blood glucose under the assumption that ↑ protein >> ↑ gluconeogenesis >> ↑ blood glucose.
However, only a very small amount of protein is converted to glucose (4g glucose from 23g of protein), and furthermore, dietary protein doesn’t increase total gluconeogenesis, nor does it increase glucose levels (even in diabetics) [1] [2].  So what happens to excess protein?  As you can see from the figure below (HT Carbsane), the gluconeogenic amino acids (green) can feed into the TCA cycle, so they are already in a good position to be oxidised.  They can of course also ultimately become glucose by first being converted to oxaloacetate, but gluconeogenesis is a very energy consuming process: “the cost of gluconeogenesis was 33% of the energy content of the produced glucose” [3], which is probably why gluconeogenesis from amino acids is so low.  Steak =/= cake
Protein Leaches Calcium from Bones
This is one the arguments in the acid-base theory of osteoporosis, which is often promoted by vegetarians and the early Paleo advocates, but is poorly supported by evidence [4] [5] [6] [7].  Protein does slightly increase urinary calcium (~1.2mg per 1g of protein) [8], but protein also increases calcium absorption [9] [10], resulting in a neutral calcium balance [9] [10].  High protein diets actually decrease the fraction of calcium from bone and reduce the rate of bone turnover [10].  This is consistent with a high protein intake resulting in a decrease in parathyroid hormone, suggesting the primary effect of protein is to increase calcium absorption and that increased calcium excretion is merely a consequence of increased absorption, rather than the calcium coming from bone to buffer the acidity of protein [9].  Overall, higher protein intakes tend to be associated with increased bone mineral density and lower fracture risk [7]
The US Reduced Fat Intake Following the Release of the Dietary Guidelines
People will cite a decline in the percentage of calories from fat following the 1970’s as evidence that we have reduced our overall fat consumption, we have gone on a low fat diet and/or that low fat is responsible for the obesity epidemic.  However, while the percentage of calories from fat has decreased, the absolute fat intake has stayed the same or increased (depending on the time period and source), which is due to a larger increase in energy from carbohydrates during that period (see the table below and More Thoughts on Macronutrient Trends).
404 (16.5%)
406 (15.1%)
261 (16.9%)
283 (15.1%)
904 (36.9%)
859 (32.8%)
557 (36.1%)
616 (32.8%)
1039 (42.4%)
1283 (49.0%)
700 (45.4%)
969 (51.6%)
It’s also incorrect to stay that low fat caused the obesity epidemic as ad libitum low fat diets can lead to weight loss [12].  And surely humans aren’t so precious that a minor change in macronutrient balance is sufficient to cause widespread obesity.
Excess Carbohydrate, Particularly Fructose, is Converted to Fat (De Novo Lipogenesis)
It’s often said that once glycogen stores are full, excess carbohydrate is converted to fat and stored.  However, de novo lipogenesis is very minor in humans contributing 0.5-10.0g depending on health (healthy, obese, diabetic) and diet (control, carb overfeeding) [13] [14] [15] [16] [17] [18], with the lower end of range (~0.5-3g) representing a standard diet and the upper end of the range (5-10g) representing pretty extreme carbohydrate overfeeding (50%).  This is consistent with the fatty acid composition of adipocytes being pretty much identical to the fatty acid composition of the diet.
Fructose is more lipogenic than glucose, but that’s not saying much.  Consuming 200g of fructose over 6 hours (far more fructose than most would have in a day) results in < 6g of palmitate from DNL in the liver and in total < 10% of the fructose was converted to fat.  By comparison, in the same study < 3% of glucose was converted to fat [18].  Another group found that following an oral fructose load the vast majority of fructose is either converted to glucose in the liver then exported to other tissues (50%), stored as liver glycogen (15%), metabolised to lactate (25%), with the remaining 10% being oxidised or undergoing de novo lipogenesis [19].  Fructose =/= fat
* It’s also ironic that the many of the people who are anti-carb or anti-fructose based on de novo lipogenesis are quite comfortable consuming ~10-50x more saturated fat from the diet than they would get on a (eucaloric) high carb diet from de novo lipogenesis (not that I’m anti-saturated fat)