Sunday, September 14, 2014

Low Carb Diet Trials: Brinkworth, et al (2009)

Participants and Diets
118 people aged 18-65 with obesity and ≥ 1 other metabolic syndrome risk factor were randomly assigned to either a low carb diet or a low fat diet.  The diets included calorie restriction (~1430 for women and 1670 for men) and were matched by calories (isocaloric).  This was done to measure the weight and metabolic effects of LC and LF diets without confounding variables such as differences in calorie intake.  The groups were similar at baseline.
Diet Group
Target Pro:Fat:Carb
Low Carb
Carb <20g for week 1-8, <40g thereafter
Low Fat
SFA <10g and <8%, additional 20g carbs for week 9-52 to remain isocaloric
Both groups adhered very well to the calorie and macronutrient targets owing most likely to the high level of support in the trial and of self-monitoring (see methods)
Low Carb
Low Fat
Calorie Intake
Calorie Intake
Weeks 1-8
Weeks 9-24
Weeks 25-36
Weeks 37-52
* Unfortunately we aren’t given baseline calorie intake
Although the low carb adhered to the macronutrient and calorie targets very well, the diet wasn’t very ketogenic, perhaps due to the protein intake, which wasn’t that high at ~130g, but might have been high enough

Both groups lost a significant amount of weight, but there wasn’t a significant difference between the groups (p =0.22 for total, p = 0.14 for the completers)
Both groups had significant (and often quite impressive) improvements in blood pressure, fasting glucose, insulin, insulin sensitivity and CRP, with no difference between the groups (see table 3)
Total cholesterol, HDL-C and LDL-C increased, while triglycerides decreased in the low carb group relative to the low fat group (see table 4).  There was no significant difference in the change between groups in non-HDL-C (p = 0.09) and Apo B (p = 0.17) between the groups*
60.0% in the low carb group completed the trial, whereas 69.2% in the low fat group completed the trial.  Given the low carb diet included calorie restriction and more restrictive macronutrient targets I’m surprised at the relative lack of difference
This trial is quite unique regarding how well the participants adhered to the diets.  The weight loss and improvement in metabolic markers was far greater in this trial than others, probably due to the extra support improving adherence.  It’s interesting that the participants in this trial reported calorie intakes similar to those in other trials, but experienced roughly 3x more weight loss, perhaps suggesting that the participants in the other trials were underreporting their calorie intake (at least to a greater degree than the participants in this trial).  Although we don’t know the baseline calorie intake among the participants in this trial so we can’t be sure.

* This is a good example where LDL-C is a waste of time to calculate (Friedwald or Iranian, it doesn’t matter).  Previous research suggests the total:HDL-C ratio (or the non-HDL-C:HDL-C ratio, because it’s essentially the same thing) is the best blood lipid risk factor of CHD and that LDL particle number (which Apo B is a measure of) is a major mechanism of CHD (but not the only one).  Non-HDL-C is a better measure of Apo B than LDL-C because just looking at LDL-C ignores other non-HDL lipoproteins (that have Apo B) such as IDL and VLDL, which are arguably either the same as or worse than LDL-C.  Low carb diets tend to decrease triglycerides and increase LDL-C, with the increase in LDL-C leading to scaremongering of ‘low carb >> high fat/SFA >> LDL-C >> heart attacks’.  However, as this study illustrates well, the increase in Apo B (the whole point of calculating LDL-C) is quite minor.  The reason why this is the case is because low carb diets tend to decrease triglycerides, which with all things other being equal results in higher calculated LDL-C, leading to an overestimation of CHD risk.


  1. Dear Steven:

    Great post! One of my favorite studies. :-)

    I have some comments about this one comment of yours:

    > "Although the low carb adhered to the macronutrient and calorie targets very well, the diet wasn’t very ketogenic, perhaps due to the protein intake, which wasn’t that high at ~130g, but might have been high enough"


    1. During the first few weeks the low-carb group went from almost 0.5 mmol/K down to about 0.3 mmol/L of β-hydroxybutyrate. According to Volek & Phinney (, the range of blood β-hydroxybutyrate in nutritional ketosis is from 0.5–5 mmol/L. (Although they don't say why they think this is the range, so perhaps 0.3 is as good as 0.5.)

    2. There was a clear difference in β-hydroxybutyrate levels between the low-carb and low-fat groups during weeks 1–8, which then basically goes away after week 8.

    3. Week 8 is when the low-carb group was instructed to go from < 20g/d of carbs to < 40 g/d of carbs.

    4. According to the food records, the low-carb group went from average of 20 g/d of carbs to average of 30 g/d at that time.

    5. Check out the companion article Figure 2, and think to yourself as you're looking at all those nine graphs that the slope of the low-carb line from baseline to week 8 is interestingly different than the slope of the low-fat line from baseline to week 8, and also interestingly different than the slope of the low-carb line on the subsequent weeks.

    6. Do you have some good references for the idea that excess protein inhibits ketogenesis? My brilliant wife, Amber, investigated that and wasn't able to come up with hard evidence, but instead just a few references to the clinical experience of experts (like Volek & Phinney):

    1. Thanks Zooko

      1. I haven't read their book but I've heard that range a fair bit. I made the comment 'not very ketogenic' because it's at the lower end of that range and they were in a pretty reasonable calorie deficit, which I hear should increase ketones

      2,3,4. Good observation

      5. Interesting, what do you put that down to?

      6. No, again this is going off what I read in the blogosphere and that it makes some sense (protein>insulin and amino acids feeding into TCA cycle). That being said I've never agreed with the steak=cake thing