All calories are not equal

I was recently drawn to read a report of a clinical trial of over-feeding, conducted by Bray et al of Pennington Biomedical Research Center. This was an inpatient study looking at the effect of protein on weight gain in people eating 40% more calories than required for maintenance of body weight.

This study was brought to my attention as evidence that the number of calories is more important than the composition of those calories, and it was said that "all the extra calories were from fat". On further examination I found that it actually shows something different.

Bray took 25 weight stable healthy adults of 18-35 years with BMI of 19 to 30. Initially they were fed a diet with 15% protein, 25% fat and 60% carbohydrate to determine their calorie intake for maintenance. Both men and women are included, which adds to the variability of the data and in a small sample like this makes statistical significance more elusive.

The intervention was to add 40% to the calorie intake of each subject, on average an extra 954 kcal/d (95% CI, 884-1022 kcal/d). Using the familiar if tired "500 calories a day = 1 lb/week" would lead us to expect about 2 pounds or 0.9 kg per week of weight gain. In addition to adding extra calories the participants were randomised to diet compositions of 5, 15 and 25% of calories as protein (low, normal and high protein or LP/NP/HP).

Here we bump into the classic difficulty of nutrition science, that of changing several things at once. Bray was interested in the effect of protein within the diet in an overfeeding situation (calorie surplus) but only the NP group were overfed at the same % protein as the run-in / stabilisation diet. So one might expect some transition effects as the diet of the LP and HP groups were both more calories and different % protein. Furthermore Bray decided that "Absolute carbohydrate intake was kept constant throughout the study " which meant everyone ate the same grams of carbohydrate on their overfeeding diet as they had on the 15% protein baseline diet. To spell this out, I have calculated the typical macronutrient intakes using data from the paper :-

	Protein (g/day)		Carbs (g)	Fat (g/day)		Daily
	Baseline	Trial	g/day	Baseline	Trial	Calories
LP	85.2	47	341	62	182	3187
NP	92	139	368	71	159	3456
HP	93.3	228	373	69	110	3392
Data from JAMA. 2012;307(1):47-55. doi:10.1001/jama.2011.1918.

Electing to use fat as the variable element to compensate for the different protein levels has the somewhat unexpected result of tripling the daily fat intake of the LP group but only adding 60% to the HP group. We end up with three quite different macronutrient splits :-

Trial	P	C	F
LP	6%	43%	51%
NP	16%	43%	41%
HP	27%	44%	29%

Moving on to the results, after 8 weeks and 50,000 excess calories Bray reported 

The weight gain in the low protein diet group was 3.16 kg (95% CI, 1.88-4.44 kg), about half that of the other 2 groups (normal protein diet: 6.05 kg [95% CI, 4.84-7.26 kg]; high protein diet: 6.51 kg [95% CI, 5.23-7.79 kg]; P = .002) 

So the LP group gained 7lbs on average, or less than 1 lb/week, despite the calorie surplus averaging 924 calories per day which should, we are frequently told, lead to a weight gain of 1.8 lbs/week.

NP gained an average of 13.3 lbs or 1.66 lbs/week with a 978 cals/day surplus and HP gained 14.3 lbs or 1.79 lbs/week with 905 kcal/day. The 95% confidence intervals for the HP group do not overlap the LP group, so the biggest gainer on LP is likely to gain less than the smallest on HP.

At 1.8 lbs/week the gain of the HP group is in close agreement with the "500 cals/day = 1 lb/week" guideline. This group put on 3.2 kg / 7 lbs of fat free mass (DEXA scan) which is almost half of their weight gain and more than the whole weight gain of the LP group, who lost a small amount 0.7kg of FFM.

The average fat gain across the groups was much more uniform than weight or fat free / lean mass gains at 3.66, 3.45 and 3.44 kg with no statistical difference between the groups. In other words one pound a week or slightly less was gained on average. See the Original Table for details.

From these results we can see that all calories are not equal in terms of their effect on weight or fat gain and the 3500 cals per lb of weight or 500 cals/day = 1 lb/week "rules of thumb" were hopelessly wrong in the low protein case for weight and wrong for fat gain in all cases. We can also see that the gain of fat was practically the same in all groups, despite fat content ranging from 29 to 51% of calories. The NP group with the highest intake of fat and biggest increase in fat from baseline gained the least weight.

Bray measured resting energy expenditure and recorded significant increases in the NP and HP groups but no change in the LP group (ventilated hood indirect calorimetry).

This increase in resting energy expenditure was strongly related to protein intake (r = 0.75, P < .001)

Total energy expenditure was measured by doubly labelled water at baseline and in the last 2 weeks of the study, again the NP group so little increase but NP and HP did use more energy

 The change in total energy expenditure in the low protein group was significantly less than in the normal protein group (Tukey-Kramer P < .05; Figure 5). There also was a positive correlation between the change in total energy expenditure and protein intake (r = 0.56, P = .004). Baseline physical activity level (calculated as total energy expenditure divided by resting energy expenditure) was low (mean [SD], 1.46 [0.28]) and did not change during the overeating period (P = .38; Figure 5). Non–resting energy expenditure was about one-third of the total energy expenditure at baseline and did not change significantly with overeating.

 Additional energy expenditure did not account for any of the additional intake in the LP group, but did account for about half the additional in the other two groups. In other words at the two higher protein intake levels half of the extra calorie intake was used up without a change in physical activity level as defined by the ratio total to resting energy expenditure.

I have attempted a balance sheet for the additional energy intake :-

	Added fat % added cals	Extra cals/day	Added fat lbs/day	Added FFM lbs/day	Calories stored per day	Stored % extra	Extra Energy use (DLW)	Missing calories per day	Extra resting energy
LP	111%	925	0.14	-0.03	595	64.40%	42	287	-21
NP	105%	978	0.14	0.11	624	63.80%	523	-169	160
HP	62%	905	0.14	0.13	628	69.40%	454	-177	227

The first column shows the calories from additional fat added to the diet as a % of the extra calories, this illustrates that the HP group got 38% of their extra calories from added protein and the LP group got all their extra calories as fat plus some more fat to compensate for reduction in protein. Keeping the carbohydrates grams/day constant throughout meant that even the NP group needed a bit of extra fat to compensate for a reduced % of energy from carbs despite the same protein as the baseline diet.

Using 9000 kcals/kg for stored fat and 1000 kcals/kg for stored lean body mass (FFM) as in the paper produces the "Calories stored per day" in column 6. This is about 2/3 of the extra calories eaten. The "missing calories" are those not accounted for in the body weight gains or the additional energy expenditure by the DLW method.

It is possible that Diet Induced Thermogenesis (DIT) is involved in the "missing calories". Westerterp found that DIT of a stable weight balanced diet was about 10% of the energy consumed, and that this value would increase by about 0.2% per 1% increase in protein content. Using these values I calculate an additional DIT over baseline of 35, 98 and 165 kcal/day in LP,NP and HP respectively - I don't know if this would be accounted for correctly by the DLW technique but the magnitude is too small on LP and with NP and HP we have accounted for more surplus calories than were fed ie DIT goes the wrong way. Bray attributes some of the extra energy use to extra protein metabolism -  

The high total energy expenditure probably reflects the higher cost of protein turnover and storage.

or in the LP case to the metabolism of fat storage -

 With the low protein diet, more than 90% of the extra energy was stored as fat. Because there was no change in lean body mass, the 6.6% increase in total energy expenditure reflects the energy cost of storing fat and is close to the estimate of 4% to 8% for fat storage derived by Flatt.31

The wording here is slightly misleading, in my opinion, as I calculate that 63.6% of the extra food intake was stored as body fat at 9000 kcal/kg in the LP group, not >90%.

Conclusions.

1. Additional calories did not turn into additional body mass at a uniform rate equivalent to 3500 calories per pound and mass was not gained at 1 pound per week per 500 calories/day difference in all groups.
2. Composition of diet counts - the low protein diet gave significantly different body composition and energy expenditure outcomes despite the same level of additional calories.
3. Despite the well controlled and measured nature of the study there are significant unaccounted calorie losses / gains when comparing the energy balance with fat and lean mass gains.
4. Only 2/3 of the additional calorie intake appears as stored fat or fat free body mass.
5. Average daily fat accumulation was approximately the same across the groups, despite significant variations in the fat content of the diets, with the LP group eating 70% more fat than HP.

Reference :-

Effect of Dietary Protein Content on Weight Gain, Energy Expenditure, and Body Composition During Overeating - A Randomized Controlled Trial

George A. Bray, MD; Steven R. Smith, MD; Lilian de Jonge, PhD; Hui Xie, PhD; Jennifer Rood, PhD; Corby K. Martin, PhD; Marlene Most, PhD; Courtney Brock, MS, RD; Susan Mancuso, BSN, RN; Leanne M. Redman, PhD

JAMA. 2012;307(1):47-55. doi:10.1001/jama.2011.1918.

PubMed entry has links to open access full text versions of the paper. clinicaltrials.gov identifier: NCT00565149