Proteome and Metabolome — Reading What Is Happening Now

A patient came in last week with a direct-to-consumer longevity panel. It said her liver age was 66, she's 52 chronologically, and she wanted to know if she has liver disease. I had no idea how to read it.

That report is making a state claim, not a trait claim. Genetic results — a risk score, a pathogenic variant — are fixed at conception and you read them once. A protein or metabolite measurement reflects what's happening in the sample at the moment it was drawn.

What does that distinction actually change?

It means the result is dynamic. Proteins are synthesized and degraded continuously — half-lives range from minutes to days. Metabolites, the small molecules: sugars, amino acids, organic acids, turn over in seconds to minutes. Last night's dinner is in that metabolome.

So the liver age number — how is it generated?

Typically a machine-learning model trained on a plasma protein panel, often SomaScan or the Olink Explore platform, predicting chronological age from protein levels, then reporting deviation as organ-specific aging. The underlying science is real. The organ-specificity claim is not validated to clinical diagnostic standards.

Meaning I can't use it to rule in or out liver disease?

Correct. There are no prospective outcome data supporting those organ-age scores as diagnostic thresholds. You'd still need ALT, AST, GGT, a FIB-4 score, and if indicated, elastography. The colored bar chart doesn't replace that workup.

What about ordering these panels myself as part of a longevity workup — do they add anything beyond apoB, Lp(a), cystatin C, fasting insulin?

Mechanistically, yes — broad protein panels identify inflammation signatures, growth factor dysregulation, things your standard lipid panel misses entirely. Clinically, we don't yet have randomized trial evidence that acting on those signals improves hard endpoints.

So what's the honest answer to the patient?

Her liver age of 66 is a signal worth noting, not a diagnosis. You evaluate it with validated hepatic tests. If those are normal, you reassure her appropriately. If they're not, you already have actionable data without the longevity panel.

And metabolomics — same situation?

Similar evidence gap for clinical decision-making. TMAO, branched-chain amino acids, acylcarnitines — associative data in cohort studies, no intervention trials demonstrating that correcting the metabolite level changes outcomes. Marketed without that evidence in most consumer contexts.

If one of these panels shows something alarming, do I act on it?

You use it to ask a question, then answer that question with a validated test. The panel tells you where to look. It does not tell you what you found.

What's the one thing you'd want me to carry out of this?

Proteome and metabolome measure present state, not fixed risk — which means they're informative and unstable. A single snapshot warrants a validated confirmatory test before any clinical decision. Don't let a colored bar chart substitute for a workup you know how to interpret.