Why we test ninety biomarkers—and why "normal" isn't enough.

She had seen four physicians in three years. She brought the labs to her first visit with me in a folder, and every page had the same word written across it in one form or another: normal. Her thyroid was normal. Her cholesterol was normal. Her blood sugar was normal. She was forty-four years old, sleeping nine hours and waking exhausted, and she had begun to wonder whether the problem was her character.

It was not her character. It was the questions the tests had been asked to answer.

A standard panel is a good instrument built for a specific job: finding disease that has already arrived. It is not built to find the slow drift that precedes disease by ten or fifteen years, and it was never designed to tell you how well you are functioning today. Those are different questions, and they need different instruments.

What a reference range actually is

This is the part almost nobody explains, and it changes how you read every lab report you will ever receive.

A reference range is not a target. It is a description of a population. Labs assemble a sample of people presumed healthy, measure the marker, and define "normal" as roughly the middle 95% of the results—discarding the extreme 2.5% at each end. The range describes where most people fall. It does not describe where you function best, and it makes no claim to.

Two consequences follow, and both matter.

The first is that the reference population is not an aspirational one. It is drawn from a country where most adults carry excess weight and a majority have some degree of metabolic dysfunction. Being statistically ordinary in that population is a low bar. A value can sit comfortably inside the range and still be a long way from where your physiology would prefer to operate.

The second is that ranges are wide, and the edges are meaningful. A marker that has moved from the bottom of its range to the top has changed enormously—often by a factor of two or three—while never once being flagged. The report says normal on every visit. The trajectory says something else entirely, and nobody is reading the trajectory, because nobody kept the earlier numbers side by side.

A reference range tells you how you compare to other people. It does not tell you how you compare to yourself five years ago.

None of this makes the reference range wrong. It makes it a screening tool being asked to do a job it was not designed for. When a patient says "my labs were fine but I feel terrible," those two facts are usually both true at once.

Why ApoB tells a different story than LDL

The clearest illustration of the gap between normal and optimal is in the lipid panel, which is also the panel most people have had run the most times.

A standard panel gives you total cholesterol, HDL, triglycerides, and LDL cholesterol—the last usually calculated rather than measured directly. LDL-C answers the question: how much cholesterol is being carried inside your LDL particles? It is a measure of cargo.

But atherosclerosis is not caused by cholesterol floating loose in the blood. It begins when a particle small enough to cross the arterial wall lodges in it and is retained. What matters is therefore not how much cholesterol is being carried, but how many particles are doing the carrying—how many opportunities per day exist for one of them to enter the wall and stay there.

ApoB measures exactly that. Every atherogenic particle—LDL, VLDL, IDL, Lp(a)—carries precisely one apolipoprotein B molecule. Count the ApoB, and you have counted the particles.

Usually the two track together, and either would tell you the same story. But in a meaningful minority of people they diverge, and this is called discordance. Someone can have many small, cholesterol-poor LDL particles: the calculated LDL-C looks reassuring because each particle carries little cargo, while the particle count—and the risk that travels with it—is high. Discordance is more common in people with insulin resistance, elevated triglycerides, or low HDL, which describes a great many people who have been told their cholesterol is fine.

When LDL-C and ApoB disagree, the accumulated evidence indicates ApoB is the better predictor of cardiovascular events. Which means a normal LDL-C, on its own, is not the reassurance it appears to be.

The related marker is Lp(a)—lipoprotein little a. It is an LDL-like particle with an extra protein attached, it is substantially inherited, and it is an independent risk factor for cardiovascular disease. Roughly one in five people carries an elevated level. It is stable across life, which means it usually needs measuring only once. It is also almost never ordered.

That last combination is what I find hard to accept: a meaningful, inherited, one-time-measurable risk factor present in twenty percent of the population, and most people reach middle age without ever having it checked. Knowing changes how aggressively everything else is managed. Not knowing changes nothing, because you cannot act on a number you don't have.

The metabolic markers that come years earlier

Fasting glucose is a late signal. By the time it rises, the compensations have usually been running for a long time.

Here is the sequence. Cells become less responsive to insulin. The pancreas compensates by producing more of it. That extra insulin succeeds in holding glucose steady—for years, sometimes a decade or more. Throughout that period, fasting glucose reads normal, because the system is working hard to keep it normal. Glucose only rises when the compensation finally fails.

So a normal fasting glucose tells you the system is currently coping. It does not tell you what it is costing.

Fasting insulin tells you the cost, and it is not on a standard panel. Measured alongside glucose it yields HOMA-IR, an index of insulin resistance that moves years before glucose does. This is the difference between finding out early, when the trajectory is highly modifiable, and finding out at the point where a diagnosis is being handed to you.

We also look at the triglyceride-to-HDL ratio, a quiet and underrated signal of metabolic health that can be calculated from a panel most people already have, and at HbA1c—useful as a three-month average, with the caveat that it reflects red blood cell lifespan and can read misleadingly in anemia or unusual cell turnover. Any single marker can mislead. The pattern across several rarely does.

Inflammation, and why one number isn't enough

hsCRP—high-sensitivity C-reactive protein—measures low-grade systemic inflammation, the kind that produces no symptoms while contributing to cardiovascular and metabolic disease over decades. It is inexpensive, widely available, and rarely run outside cardiology.

It also needs interpreting rather than simply reading. hsCRP rises with any acute illness, so a single elevated value in someone who had a cold last week means little. It is a marker that only becomes informative when you have more than one measurement and can see whether the elevation persists. This is true of a great deal of what we test, and it is the reason a biomarker panel is a starting point rather than a verdict.

Thyroid: why TSH alone answers half the question

Most thyroid screening is a single marker: TSH. But TSH is not a thyroid hormone. It is a pituitary hormone—the brain's instruction to the thyroid. Measuring TSH tells you how loudly the brain is asking. It does not tell you what the thyroid produced, or whether the body converted it into the form that actually does the work.

Thyroid hormone is produced mostly as T4, which is comparatively inactive and must be converted to T3. That conversion happens in peripheral tissue and is affected by stress, illness, inflammation, calorie restriction and nutrient status. A person can have a perfectly acceptable TSH, adequate T4, poor conversion, and symptoms that map cleanly onto low thyroid function.

So we look at free T4, free T3, and where the picture warrants it, reverse T3 and thyroid antibodies—TPO and thyroglobulin. Antibodies in particular can be present for years before TSH moves at all, which means autoimmune thyroid disease is often detectable well before it is diagnosed.

Hormones: total, free, and the protein in between

Testosterone is usually reported as a total: everything circulating in the blood. But most of it is bound to sex hormone-binding globulin and albumin, and bound hormone is not available to tissue. What your body can actually use is the free fraction.

SHBG rises with age, with thyroid changes, with certain medications, and with some patterns of insulin sensitivity. When it rises, total testosterone can hold perfectly steady while free testosterone falls—and the person feels the difference long before any number is flagged. Measuring total alone will miss this entirely.

The same logic applies across the endocrine picture: estradiol, progesterone, DHEA-S, LH, FSH, and IGF-1 as a proxy for growth hormone. Individually each is a fragment. Read together, and read against the person in front of me rather than against a population average, they describe a system.

The nutrient markers that explain symptoms

Some of the most common complaints I see—fatigue, brain fog, hair shedding, poor recovery—resolve at least partly at the nutrient level, and the relevant markers are rarely measured.

Ferritin is a good example of why interpretation matters more than the number. It reflects iron storage, and low-normal ferritin is a well-recognised cause of fatigue and hair loss in menstruating women, well above the threshold for anemia. But ferritin is also an acute phase reactant: it rises with inflammation. A normal ferritin in someone with elevated hsCRP may be masking genuine iron deficiency. You cannot read one without the other.

We also look at vitamin D, B12 and folate, magnesium, zinc, and homocysteine—which sits at the junction of B-vitamin status and cardiovascular risk and is one of the more useful single markers we run.

Almost nothing on a comprehensive panel means anything on its own. The interpretation is the medicine.

Why we test again

A single panel is a photograph. What we are actually after is the film.

Because reference ranges are wide, the most valuable information is often not where a marker sits but which direction it is moving and how fast. A fasting insulin that has climbed steadily for three years is telling you something urgent, and every one of those values may have been reported as normal. You only see it if someone laid the results side by side.

This is also the honest answer to whether an intervention worked. Feeling better matters enormously and I do not discount it—but feeling better is confounded by season, sleep, stress, expectation, and the simple fact of being taken seriously. Re-testing separates what is actually changing in your biology from what changed in your week. Sometimes the answer is that a protocol is not doing what we hoped, and we would rather know that at three months than at three years.

What "normal isn't enough" actually means

I want to be careful here, because this phrase gets used to sell things.

It does not mean every marker should be driven to some theoretical ideal. It does not mean more testing is automatically better medicine—an unnecessary test that produces an ambiguous result leads to anxiety and more tests, and that is a real harm. And it does not mean your previous clinicians were negligent. A fifteen-minute visit built around ruling out disease is doing what the system asked of it.

What it means is narrower and, I think, more defensible. When someone feels unwell and their standard panel is unremarkable, "normal" is not a conclusion. It is the end of what that particular instrument can see. There is a great deal more that can be measured, much of it inexpensive and widely available, and quite often it explains what the first panel could not.

It also means treating the numbers as one input among several. A panel does not know that you have been sleeping four hours, or grieving, or training for something. I have seen results that looked alarming resolve completely once the context was known, and results that looked unremarkable turn out to be the thread worth pulling. The data narrows the possibilities. The conversation decides which one is yours.

The patient in the folder, in the end, had three things: iron stores near the bottom of the range, a thyroid conversion problem invisible to TSH, and a fasting insulin that had been climbing for years. Not one of them had ever been flagged. Not one of them was exotic or expensive to measure. They simply had not been asked about.

That is the whole argument for testing ninety markers instead of twelve. Not that more is better. That the right questions get asked while the answers still matter.

This article is educational and is not medical advice, diagnosis, or treatment. Laboratory results require interpretation in the context of your history, symptoms, medications, and examination, and optimal values differ from person to person. Nothing here should be used to start, stop, or change any treatment. Please discuss your own results with a qualified clinician who knows your case.