Removing genetic effects from plasma proteins is an effective strategy to increase power for biomarker discovery and clinical trial design, consistent with the largely non-causal role of most plasma proteins in disease risk.

👏 Lead by Daniela Fusco and Zhiyu Yang

✅ 88% of 1,300+ protein–disease pairs showed stronger associations after genetic adjustment
✅ Equivalent to needing 30% fewer samples for the same power
✅ Multi-protein scores improved prediction for 7 major diseases
✅ Adjusted proteins aligned more closely with exposome

Plasma proteins reflect both genes and environment.

But when genetic variation unrelated to disease drives protein levels, it can dilute biomarker signals.

We genetically-adjust 94 highly heritable proteins in ~40K UK Biobank individuals to see if removing genetic effects helps

Link to the pre-print: www.medrxiv.org/content/10.1...

A fantastic collaboration with
@finngen.bsky.social &
Finnish Clinical Biobank Tampere, led by Rodos Rodosthenous & Leena Viiri.

I personally learned a lot about running an RCT: we need to make them simpler!

@fimm-uh.bsky.social
@hilife-helsinki.bsky.social

Takeaway:

While genetics strongly predict body weight at baseline, they do not determine who benefits from dietary coaching. In other words, behavioral interventions can overcome genetic risk in non-diabetic overweight and mildly obese adults.

📊 Results:

✅ Excellent retention: 90% participants returned at 6 months (in both trial arms)
✅ Diet worked: intervention group lost ~5% body weight vs controls
❌ But… the effectiveness of the intervention did not differ between those with high vs. low genetic risk for higher BMI.

🧠 What makes GENEROOS unique?

👉 It’s the first prospective RCT to directly test this hypotesis (others did so retrospectively) and the first to recruit participants from the extreme tails (top & bottom 5%) of the BMI polygenic score.

223 non-diabetic adults (BMI 23–36 kg/m²) took part.

Thanks to the many biobanks involved!
And particularly Zhiyu Yang for leading this!

Why it matters? progression genetics is more relevant drug targets + clinical care.

👉 Solutions:
• Build larger, harmonized cohorts & refined progression phenotypes
• Use proxy phenotypes from general population

Read: www.nature.com/articles/s41...

comparing to one clinical prediction model is one way to think about this. the other is about the re-use of generated trajectories for multiple purposes in just one go: readmission rate, mortality, healthcare costs, disease risk and so on…

Thanks Michel! let’s chat. we want to have this pan-european.