Running the Longevity Workup in the Right Order

I had a patient hand me a sixty-page longevity report last week. Forty flagged values, a biological age, a polygenic score — which is a score derived from hundreds of common genetic variants — and zero guidance on what to do. Where do I even start?

You start by asking whether the order was right before a single result was run. These reports fail when they're assembled as a catalog. The sequence is the clinical logic.

What's the correct sequence?

Three layers. Mortality first — blood pressure toward 120 systolic, apoB driven low on a lifetime-exposure model, colonoscopy at 45, smoking. Then morbidity: cardiovascular, metabolic, neurodegenerative, cancer risk. Then performance: how fast function is declining. Nothing from an upper layer gets acted on while a lower-layer problem is active and uncontrolled.

So if someone's apoB is 140 and untreated, I'm not pivoting to their microbiome data.

Correct. Guideline targets override optimization targets. That's not a soft rule — it's the rule that keeps the workup from becoming theater.

What about the genome? People always want that first.

Germline sequencing — reading the DNA you were born with — is standard for BRCA1, BRCA2, Lynch syndrome genes, and FH if family history or LDL supports it. The polygenic scores for cardiovascular disease have trial-level association data but no randomized evidence yet that acting on the score improves outcomes beyond acting on the lipids themselves. That's a tracked item, not a standard one.

How confident are you in the biological age outputs — the ones based on methylation patterns in DNA?

Epidemiological confidence is real — accelerated epigenetic age associates with mortality in large cohorts. Clinical confidence is low. No intervention trial has used an epigenetic clock as a primary endpoint with hard outcomes. It's a process measure worth a baseline and a dated repeat, disclosed as preliminary.

What about proteomics — measuring thousands of proteins in blood simultaneously?

The SomaScan and Olink platforms have real signal for disease risk, but no randomized trial has shown that acting on a proteomic panel changes outcomes. For some patients as a research-adjacent baseline — with that said out loud first.

Imaging — coronary artery calcium, whole-body MRI?

CAC scoring is guideline-supported by ACC/AHA for intermediate-risk patients to guide statin decisions. Whole-body MRI is standard with disclosure — incidental findings requiring workup occur in roughly a third of scans, and the patient needs to understand that before the magnet turns on.

So I have a patient with forty flagged values. How do I actually find the one first action?

Sort everything by layer. Find the lowest-layer abnormality that's both actionable and has the highest evidence grade. That's Monday's action. Everything else gets a date and a category — standard, for some patients, with disclosure, or tracked — and you tell the patient exactly which bucket each item is in.

That's a very different conversation than handing them the report.

The report is raw data. The clinical task is converting it into a ranked plan with stated confidence on each element. If you can't say why something is in the plan and what evidence grade it carries, it shouldn't be in the plan.

What's the one thing you'd want me to hold onto?

The order is the clinical logic. A test that's only interpretable in the light of a layer below it has no business being the first thing you act on. Run the layers in sequence, state your confidence, find the lowest actionable abnormality, and that's Monday.