Energy-balance studies reveal associations between gut microbes, caloric load, and nutrient absorption in humans
Jumpertz et al. held 21 men (12 lean, 9 obese) on a metabolic ward, fed each a 2400- and a 3400-kcal/d diet for 3 days in random order, and measured every calorie in, in stool and in urine by bomb calorimetry, while sequencing the stool bacteria.
Stool energy loss did not differ between lean and obese men eating the same food: 4.9 ± 1.8% vs 4.8 ± 1.4% of intake at 2400 kcal/d, and 3.8 ± 1.1% vs 4.6 ± 1.8% at 3400 kcal/d. Between individuals it ranged from 2.1 to 9.2% at 2400 kcal/d.
In the LEAN men only, a shift of the stool bacteria toward Firmicutes was associated with less energy lost: a 20% increase in Firmicutes went with about 150 kcal more absorbed. The association was not seen in the obese men. On the way the authors found that the food itself carried more energy than its product labels stated.
Protocol
21 adult white men, non-smokers, healthy: 12 lean (BMI 23.4 ± 1.7; one overweight man with BMI 26.1 included) and 9 obese (BMI 40.4 ± 4.6); aged 32.8 ± 9.2 vs 35.8 ± 10.6 y. Screened for occult blood, parasites and coeliac disease; no antibiotics or probiotics within 3 months.
Inpatient: 3 days of weight-maintaining diet, then 2400 and 3400 kcal/d diets for 3 days each in random order, each followed by a 3-day washout on the weight-maintaining diet. All diets ~24% protein, 16% fat, 60% carbohydrate, similar fibre.
Duplicate trays of every meal analysed by bomb calorimetry; uneaten food weighed back and analysed. Stool collected between two blue dye markers; urine collected over the diet days; all samples lyophilised and burned in a bomb calorimeter.
Stool bacteria by pyrosequencing of 16S rRNA genes before and during each diet. Nutrient load expressed as % of each man's weight-maintaining energy needs (%WMEN).
Key Results
| Outcome | Lean | Obese | Comparison (full text) |
|---|---|---|---|
| Stool energy, 2400 kcal/d | 134.3 ± 48.9 kcal/d; 4.9 ± 1.8% of intake | 133.2 ± 44.8 kcal/d; 4.8 ± 1.4% | P = 0.96 (kcal); P = 0.87 (%) |
| Stool energy, 3400 kcal/d | 145.1 ± 42.7 kcal/d; 3.8 ± 1.1% | 173.7 ± 65.0 kcal/d; 4.6 ± 1.8% | P = 0.21; P = 0.240 |
| Range between individuals (% of intake lost in stool) | — | — | 2.1–9.2% at 2400 kcal/d; 1.6–7.6% at 3400 kcal/d |
| Change in stool energy, 3400 vs 2400 | −1.3 ± 1.9% (P = 0.04) | −0.2 ± 1.2% (P = 0.59) | lean men lost relatively less when overfed |
| Urine energy | 3.2 ± 1.1% (2400); 2.8 ± 0.9% (3400) | 3.5 ± 1.1%; 2.9 ± 0.7% | not different |
| Microbiota vs stool energy | 20% more Firmicutes ≈ 150 kcal more absorbed (20% more Bacteroidetes ≈ 150 kcal less) | not observed (Bacteroidetes P = 0.14, Firmicutes P = 0.15) | interaction with body-weight status P = 0.02 |
| Measured vs labelled meal energy | — | — | bomb calorimetry of duplicate meals higher and more varied than the product labels (P < 0.0001; values shown only in a figure) |
| Gut transit time | 1139 ± 391 min (2400) | 951 ± 407 min | P = 0.33; not related to stool calories |
Mechanisms
Measured: the share of stool bacteria from the two dominant phyla changed within 3 days of a change in nutrient load (with 3400 kcal/d, Firmicutes r = 0.47, P = 0.04 against %WMEN).
Measured (association): in lean men, phylum shifts tracked stool calories (Bacteroidetes r = 0.52, Firmicutes r = −0.50); not in obese men.
Authors' interpretation: gut microbes may have a role in regulating nutrient harvest, and the degree of overfeeding relative to each man's needs may explain why the association was clearer in lean men. The study cannot show cause.
Inference, not tested: the ≈150 kcal figure is the slope of an association scaled to a 20% shift in Firmicutes (observed changes ranged −8% to 34%), not an amount any individual was shown to gain.
Practical Application
The same food gives different people only modestly different energy: with identical meals, lean and obese men lost the same share of intake in stool (4.9 ± 1.8% vs 4.8 ± 1.4% at 2400 kcal/d; 21 men, B). Individuals ranged from 2.1 to 9.2% of intake, which at 2400 kcal/d is roughly 50 to 220 kcal/d (arithmetic). The widely quoted microbiota effect of about 150 kcal was an association in the lean men only, over 3 days, not something shown to cause a difference.
Labels can be off in the other direction too: the hospital meals carried more energy by bomb calorimetry than their product labels stated (the size is in a figure, not the text). A food log is an estimate either way.
Absorption becomes slightly more efficient with a big surplus in lean men (stool loss −1.3 ± 1.9% of intake at 3400 vs 2400 kcal/d), which reduces, not increases, the "the body wastes the extra" hope.
Not a basis for gut tests or "harvest-reducing" probiotics: nothing here manipulated the microbes (SYNTH-87).
Limitations
21 men only, white, non-athletic; 3 days per diet; the obese group was severely obese (BMI 40.4).
The microbiota finding is correlational and appeared in one subgroup; there was also a possible diet-order effect on the microbiota association.
Planned 2400 and 3400 kcal diets carried more energy than labelled, so loss had to be expressed as a percentage of measured intake.
Urine collection was not marker-timed like stool.
16S sequencing at phylum level (2011 methods); species and functions not measured.
How this entry was checked and changed
Why this level
the diets were randomised, but the microbiota-absorption finding is correlational, in n = 21 men (12 lean, 9 obese) over 3 days per diet; read from the full text
Full text (checked 2026-10-13)
Twelve lean and 9 obese adult white men; 2400- or 3400-kcal/d diet for 3 d in random crossover order, each followed by a 3-d washout (full text).
Stool energy at 2400 kcal/d: 134.3 ± 48.9 vs 133.2 ± 44.8 kcal/d (P = 0.96); 4.9 ± 1.8% vs 4.8 ± 1.4% (P = 0.87) (full text).
Stool energy at 3400 kcal/d: 145.1 ± 42.7 vs 173.7 ± 65.0 kcal/d (P = 0.21); 3.8 ± 1.1% vs 4.6 ± 1.8% (P = 0.240) (full text).
Interindividual range of calories lost in stools: 2.1–9.2% (2400 kcal/d) and 1.6–7.6% (3400 kcal/d) (full text).
In lean men a 20% increase in Firmicutes was associated with an increase in nutrient absorption of ≈150 kcal; not observed in obese men (P = 0.14, P = 0.15); interaction P = 0.02 (full text).
Lean men: change in stool energy loss with 3400 vs 2400 kcal/d −1.3 ± 1.9% (P = 0.04); obese −0.2 ± 1.2% (P = 0.59) (full text).
Measured calorie content of each diet "was higher and more varied than indicated by the food labels" (full text).