The men tended to be older (38 vs 33) and less obese (34 vs 38 BMI) with p-value 0.15 or lower but not reaching statistical significance (9 women and 10 men is a modest sample size).
As the women did not show a statistically significant measured weight loss I'm going to be overtly sexist and concetrate on the men. This isn't just self interest as they're a lot younger than me, much more insulin sensitive and less diabetic and currently I am on the verge of 25 BMI.
What results did the men achieve ? A measured weight loss of 2.1 kg on restricted carbohydrate (RC) was better than the 1.25 kg recorded on restricted fat (RF), p=0.014. Measured fat loss (DEXA) was not statistically different (p=0.8) between the diets with RC losing 0.82 kg and RF 0.78 kg.
This suggests weight loss was 39% fat on RC and 63% fat on RF, as glycogen / water was being depleted on RC. Another way of looking at this is that the calorie restriction required for 1 lb of weight loss was 975 on RC and 1975 on RF. I calculated this from the total weight loss and daily calorie deficit reported (-748 vs -906 respectively) and it suggest glycogen depletion is at work in both cases - the figure 3B below shows the results of CHO balance calculation for men & women combined :-
The red dots show carbohydrate depletion in restricted fat well below the red line of the model prediction and the blue dots of restricted carbs point to a large but declining carbohydrate depletion, again below the blue line prediction.
Looking through the supplementary data tables the men on restricted carbohydrate had a significant change in total daily energy expenditure (TDEE) by 139 calories (+/- 36 SEM). Much of this is from a reduction in Sleeping metabolic rate (SMR) of 98 calories (+/- 46, p=0.07).
Digging more into the substrate balance, the supplementary data reveals that average fat oxidation increased by 393 cals/day on restricted carbohydrate. This is only half the 800 calorie reduction in carb intake :
Here the "CHO oxidation" is the difference between the recorded fall in oxidation and the reduced CHO intake - we may have expected an 800 calorie reduction in CHO oxidation to match the intake but only saw a reduction of 605 due to depletion and oxidation of carbohydrate reserves. This effectively suppresses fat oxidation, as does the reduction in energy expenditure.
A longer term experiment, or one with a 2-3 week run-in on restricted carbs, is likely to show greater fat oxidation as the CHO reserves stabilise and possibly a restoration of TDEE as the body adapts to its new fuelling regime.
Other significant changes (P<0.10) in the restricted carb diet were in marginally increased ketones (B-OHB), reduced total and HDL cholesterol, reduced triglycerides, reduced leptin and ghrelin.
The restricted fat diet recorded a reduction in fasting blood glucose & insulin, reduced HOMA-IR, lower leptin and reductions in both HDL & total cholesterol. The HDL reduction of RF was double that of RC. Comparison of the two diets showed significant differences (P<0.05) only in ketones, HDL and ghrelin.
The Insulin Hypothesis.
One of the studies goals was to explore the "Insulin Hypothesis" of Gary Taubes et al. Clearly there was weight loss in the fat reduced diet where carbohydrate intake was maintained. In the men the fasting insulin fell on the RF diet by 2.6 pg/ml, p=0.033, from a baseline of 10. The insulin secretion measured by 24h C-peptide is reported to have decreased by 22% (p=0.001) in the case of the RC diet which is perhaps an illustration of how reduced insulin secretion is associated with increased fat oxidation - the RC diet oxidised 400 kcal/day more fat where the RF diet saw a very small NS increase. Less carbs (A), less insulin (B), more fat oxidation (G) - in line with what Taubes proposes :-
If the alternative to the Insulin Hypothesis is "a calorie is a calorie" then the latter hypothesis didn't come out to well in this study either. If the effect of calories is uniform regardless of composition how does one explain Fig 2D & 2G where an identical change in energy intake produces a radically different fuel oxidation pattern ?