What the research found

The DIRECT PLUS trial followed 294 adults with metabolic risk factors (abdominal obesity or dyslipidemia) for 18 months, comparing three dietary approaches: standard healthy guidelines, a conventional Mediterranean diet, and a "green" Mediterranean diet enriched with polyphenols. All groups received gym access and activity support; the intervention isolated diet quality rather than just activity level.

The green Mediterranean variant added concentrated polyphenol sources—3–4 cups of green tea daily, a Mankai shake (a protein-rich aquatic plant), and walnuts—totaling roughly 1,240 mg of polyphenols per day, about double the standard Mediterranean arm. Compared to both control and standard Mediterranean groups, the high-polyphenol approach showed measurable shifts in visceral fat loss (14.1% versus 6.0% and 4.2% respectively), aortic wall elasticity (15% improvement in stiffness versus 7.3% and 4.8%), and, in participants aged 50+, markers of brain volume preservation. Changes in cholesterol and markers of polyphenol metabolism (urolithin-A and tyrosol) correlated with vascular and neurological benefits.

The trial was 18 months long and included measurements via MRI, blood work, and epigenetic analysis. Notably, the cohort was 88% male, mean age 51, and employed at a single research facility in Israel—a relatively homogeneous population that may not generalize to other demographics.

Why it matters for you

If you're tracking lipid biomarkers or epigenetic aging clocks like TruAge, the green Mediterranean approach offers a testable dietary lever beyond standard nutrition advice. The DIRECT PLUS findings suggest that polyphenol density—not just calorie restriction or moving more—correlates with changes in LDL and total cholesterol alongside improvements in vascular elasticity. For users already on lipid-lowering interventions (medications, GLP-1 agonists, or training), this represents a potential stacking opportunity to close remaining gaps in lipid panels.

The brain atrophy data is particularly relevant if you're monitoring cognitive aging or neurodegenerative risk. The trial found that insulin sensitivity improvements and specific polyphenol metabolites tracked with preserved hippocampal volume and reduced ventricular expansion in older participants. This suggests the mechanism isn't just "eat healthy food," but rather that the polyphenol load and its effect on glucose regulation may drive measurable neuroprotection—something you could monitor via repeat MRI or blood biomarkers of insulin resistance.

Practically, this means adding high-polyphenol foods (green tea, walnuts, Mankai if accessible) to your existing protocol rather than overhauling your diet entirely. The trial controlled for calories and activity, so the specificity of which foods you choose appears to matter. If you retest biomarkers every 6–12 weeks (lipids, fasting glucose, HbA1c), you'd be able to see whether these additions move the metrics you care about, similar to how you'd assess any other intervention.

Caveats

• Population mismatch: 88% male, mean age 51, single workplace location in Israel. Mechanisms are likely sex-agnostic, but the trial wasn't designed to test effectiveness in women, especially perimenopausal or postmenopausal populations.
• Polyphenol metabolites, not just foods: The benefits correlated with specific urinary metabolites (urolithin-A, tyrosol) and improved insulin sensitivity, not just the presence of these foods in the diet. Individual ability to convert plant polyphenols into active metabolites varies.
• Mankai accessibility: This aquatic plant is not widely available outside Israel and Europe. Generalizing to more common polyphenol sources (tea, berries, nuts) may yield different effect sizes.
• Brain imaging limited to MRI: Only participants aged 50+ received brain MRI; younger participants in the trial were not assessed for age-related atrophy.
• Short-term follow-up: 18 months is reasonable but doesn't address long-term adherence, sustaining benefits, or whether effects persist after the trial ends.