Study snapshot
- Published
- 2025-08-28
- Study design
- Controlled 2x2 crossover feeding study comparing ultra-processed with unprocessed diets at adequate and excess calorie assignments
- Population
- Healthy men aged 20 to 35 with BMI 18.5 to 30 and semen quality above the WHO lower reference
- Sample size
- 43
- Duration
- Three weeks per diet condition with washout
- Peer reviewed
- Yes
- Main reported result
- The supplied ultra-processed diet package caused greater short-term body-weight and fat-mass gains than the unprocessed package at both assigned calorie levels.
- Reported effect
- Ultra-processed minus unprocessed fat mass: 0.998 kg (95% CI 0.405 to 1.641) at adequate calories and 0.9555 kg (95% CI 0.377 to 1.533) at excess calories.
- Important limitations
-
- Small, short, free-living study in young men only.
- The diets differed in fiber, saturated fat, added sugar, refined grains, dairy and contaminants as well as NOVA processing category.
- The report presents exploratory secondary outcomes; the registered primary sperm-DNA-methylation outcome was not reported here.
- Sperm motility changed, but semen parameters are not pregnancy or infertility outcomes.
- Funding
- Novo Nordisk Foundation grants and French public research programs.
- Conflicts of interest
- One author reported Novo Nordisk consulting, honoraria and research funding; the foundation center is supported by an unrestricted donation.
How this article was sourced
Selected from a date-audited 500-paper PubMed screening ledger, then checked against the primary paper record and supporting research before drafting.
This page explains one recent research result in context. It does not turn a study average, biomarker, or association into an individual diagnosis, treatment promise, or personal forecast.
The question calories alone could not answer
Forty-three healthy men aged 20 to 35 entered a controlled 2x2 crossover study. They received ultra-processed and unprocessed diet packages under adequate- and excess-calorie assignments, spending three weeks in each condition with washout. The crossover design lets the investigators compare how the same person responded to different menus instead of comparing habitual eaters who may differ in dozens of ways.
The study addresses a popular claim: if calorie targets match, processing should stop mattering. Assigned calories did not eliminate the observed difference. Yet the experiment remained free-living, and assignment is not identical to measured intake or absorption. Its strongest statement concerns the supplied packages at their assigned energy levels, not an abstract calorie-versus-processing contest.
The kilogram-sized result
At the adequate-calorie assignment, body weight was 1.4055 kg higher on the ultra-processed diet than on the unprocessed diet; the 95% confidence interval ran from 0.566 to 2.251 kg. Under excess calories, the corresponding difference was 1.2768 kg, with a 95% interval from 0.467 to 2.083 kg.
Fat mass accounted for much of that contrast. The adequate-calorie difference was 0.998 kg, with a 95% interval from 0.405 to 1.641 kg. Under excess calories it was 0.9555 kg, with an interval from 0.377 to 1.533 kg. These are unusually concrete estimates for a three-week food study, but still short-term averages from 43 men.
View accessible chart data
| Group or point | Value | Lower bound | Upper bound | Note |
|---|---|---|---|---|
| Body weight - adequate calories | 1.406 kg | 0.566 kg | 2.251 kg | adjusted p=0.027 |
| Body weight - excess calories | 1.277 kg | 0.467 kg | 2.083 kg | adjusted p=0.032 |
| Fat mass - adequate calories | 0.998 kg | 0.405 kg | 1.641 kg | adjusted p=0.027 |
| Fat mass - excess calories | 0.956 kg | 0.377 kg | 1.533 kg | adjusted p=0.027 |
Processing was bundled with a different menu
The diets differed not only by NOVA category but also in fiber, saturated fat, added sugar, refined grains, dairy and measured contaminants. The packet does not identify which of those co-moving differences produced the result. Randomization tells us the packages differed in effect; it does not grant sole causal credit to an industrial label.
That is why broader evidence must remain separate. Recent cohort work has examined plant-based diets and ultra-processed foods, a randomized study has targeted ultra-processed-food reduction alongside weight loss in fatty liver, and a meta-analysis has reviewed adolescent obesity risk. They supply context across longer horizons and different populations, not a mechanistic explanation for this crossover result.
The fertility headline the study cannot support
Total sperm motility decreased during the ultra-processed condition, but the paper characterizes the semen evidence as a trend rather than a fertility outcome. Semen parameters can generate hypotheses; they do not count pregnancies and cannot establish infertility. Every participant began above the World Health Organization lower reference for semen quality, further narrowing generalization.
The report also focused on exploratory secondary outcomes, while the registered primary sperm-DNA-methylation outcome was not reported in this paper. That does not erase the weight and fat-mass findings. It does mean reproductive claims require extra restraint, especially when a striking biomarker can become a warning about future fatherhood without supporting clinical evidence.
Where the result lands after the excitement
The honest headline is still noteworthy: even with energy assignments, the ultra-processed package produced roughly a kilogram more fat mass over three weeks. The study does not tell a shopper which additive, texture or substitution mattered, and it does not show whether the gap would widen, shrink or reverse over months.
Funding came from Novo Nordisk Foundation grants and French public research programs. One author reported Novo Nordisk consulting, honoraria and research support, and the foundation center is backed by an unrestricted donation. Those disclosures do not decide the result, but they belong beside its exploratory status. What remains is a sharp replication target, not a verdict on every packaged food.
Replication should preserve the strongest feature - supplying clearly characterized menus at more than one energy level - while measuring actual intake and separating prespecified from exploratory outcomes. Including women and a wider age range would show whether the kilogram-sized contrast is unusually specific to young men or appears across populations.
The confidence intervals already make the experiment more informative than a bare before-and-after headline. They show uncertainty around each kilogram estimate and keep the public story tied to the comparison the study actually made.