Low Ultra-Processed Diets Beat Both Omnivore and Vegetarian Labels for Aging Metabolic Health
Cutting ultra-processed foods improved insulin sensitivity, cholesterol, and body weight in older adults — regardless of whether they ate meat or went vegetarian.
Summary
A randomized crossover trial in 36 older adults found that switching from a high ultra-processed food (UPF) diet to either a minimally processed omnivorous or lacto-ovo vegetarian diet produced nearly identical metabolic benefits. Both interventions unintentionally reduced caloric intake by 333–437 kcal/day, leading to 3.8–4.4 kg weight loss and 2.6–2.9 kg fat mass reduction. Insulin sensitivity (HOMA-IR), LDL, total cholesterol, apolipoprotein B, and CRP all improved significantly. Leptin fell and FGF21 rose in both groups, suggesting shared nutrient-sensing pathways. Crucially, when participants returned to high-UPF eating at one-year follow-up, all gains largely reversed — underscoring that UPF reduction, not dietary protein source, drives cardiometabolic improvement in aging adults.
Detailed Summary
As Americans age, cardiometabolic risk rises sharply, threatening healthspan and straining healthcare systems. Ultra-processed foods now comprise roughly 57% of the average U.S. adult's caloric intake, and their role in worsening metabolic health is increasingly recognized — yet no prior feeding trial had tested UPF reduction specifically in older adults within the framework of the Dietary Guidelines for Americans (DGA).
The PRODMED2 trial enrolled 36 community-dwelling older adults in an 8-week crossover design comparing two DGA-aligned, low-UPF diets: one centered on minimally processed pork (omnivorous) and one on lentils (lacto-ovo vegetarian). Each diet phase was preceded by measurement of a high-UPF habitual baseline (~50% of energy from UPF) and separated by a 2-week washout. A one-year follow-up assessed durability of effects.
Both low-UPF diets (~13% of energy from UPF) produced strikingly similar outcomes. Participants spontaneously reduced caloric intake, lost weight and fat mass, and showed significant improvements in HOMA-IR, fasting insulin, C-peptide, total cholesterol, LDL, apolipoprotein B, non-HDL cholesterol, and CRP. Fasting leptin declined and FGF21 increased in both groups, implicating shared nutrient-sensing and energy-balance pathways. Critically, no statistically significant differences emerged between the two diet types on any outcome.
At the one-year follow-up, UPF consumption had rebounded to ~44% of energy, and body weight, adiposity, and biomarkers largely returned toward baseline values. This reversal powerfully illustrates that the benefits were driven by sustained low-UPF eating rather than any lasting physiological adaptation.
The findings challenge the common framing of plant-based versus animal-based diets as the primary dietary lever for metabolic health in older adults. Instead, food processing level appears to be the dominant variable. The study supports low-UPF eating as a practical, flexible strategy for healthy aging that accommodates both omnivorous and vegetarian preferences.
Key Findings
- Both low-UPF diets reduced body weight by ~3.8–4.4 kg and fat mass by ~2.6–2.9 kg without calorie restriction.
- HOMA-IR, LDL, total cholesterol, apolipoprotein B, and CRP improved significantly versus high-UPF baseline in both diet groups.
- No significant metabolic differences were found between the omnivorous (pork) and vegetarian (lentil) diet arms.
- FGF21 rose and leptin fell in both groups, suggesting UPF reduction activates nutrient-sensing pathways.
- All metabolic gains largely reversed at one-year follow-up when UPF intake rebounded to 44% of energy.
Methodology
Randomized crossover feeding trial (PRODMED2, NCT05581953) in 36 community-dwelling older adults; each diet phase lasted 8 weeks without caloric restriction, separated by a 2-week washout, with a ~1-year post-intervention follow-up. Outcomes were analyzed using robust linear mixed-effects models adjusted for relevant covariates.
Study Limitations
The trial enrolled only 36 participants, limiting statistical power and generalizability. The crossover design and self-selected community sample may introduce order effects and selection bias. Only abstract-level data are available; full dietary composition details and subgroup analyses cannot be assessed.
Enjoyed this summary?
Get the latest longevity research delivered to your inbox every week.
Enter your email to subscribe: