The Metabolic Canon

The twenty-year paradox

A type 2 diabetes diagnosis feels like it took three days. The blood draw on Tuesday. The portal notification on Friday. The phone call that afternoon. The prescription. The word. All of it inside one week, and from the inside, all of it sudden.

In reality, the diagnosis took twenty years.

There is no ambulance in this story, which is part of why nobody tells it. Atherosclerosis at least ends with a dramatic minute — the chest, the sirens, the stent. Metabolic disease ends with a phone call and a word, and then it quietly goes on to write the ending of several other diseases: coronary artery disease, kidney failure, retinal damage, peripheral nerve damage, fatty liver, and, over a long enough horizon, cognitive decline. It is the least theatrical of the major diseases and one of the most consequential.

It is also the most misnamed. We call it a disease of blood sugar because blood sugar is what we happen to measure. It is more accurate to call it a disease of energy storage that eventually shows up as blood sugar — and "eventually" is doing an enormous amount of work in that sentence.

By the time the standard threshold is crossed, roughly half of the insulin-producing capacity of the pancreas is typically already gone. Not declining. Gone. That single fact should reframe everything about how this disease is screened for, because it means the moment of diagnosis is not the beginning of the disease. It is somewhere near the end of its silent phase.

This is the story of those twenty years. Where it starts. What is happening inside a muscle cell while every lab result comes back perfect. Why the number we test for is the last number to move. What the other numbers on the page were saying while nobody read them. And why the cheapest test that would have found it fifteen years early is a test almost nobody has ever been given.

We're going to start with a single molecule and a locked door.

(A note before we begin: everything here concerns type 2 diabetes and the insulin-resistant state that precedes it. Type 1 diabetes is a different disease with a different cause — an autoimmune destruction of the insulin-producing cells — and none of the chronology below applies to it.)

Glucose, restored

Sugar has been demonized for a generation. So it's worth saying clearly, before anything else:

Glucose is one of the most useful molecules in human biology. It is not optional. It is not, in itself, a problem. It is the fastest and cleanest fuel your body has — the one substrate every cell in your body can convert into ATP, adenosine triphosphate, the universal currency that every cell actually spends. Muscles contract on ATP. Nerves fire with it. Every enzyme, every pump, every repair process runs on it.

Fully oxidized in the presence of oxygen, a single molecule of glucose yields roughly thirty molecules of ATP. Broken down quickly without oxygen — the anaerobic route your muscles use in a sprint — it yields two, plus lactate. That gap is why oxygen matters so much to endurance, and why the same fuel can be burned two very different ways depending on how fast you need it.

Some tissues have no alternative at all. Red blood cells have no mitochondria, so they cannot burn fat or ketones; they run on glucose exclusively, for their entire four-month lifespan. And your brain, roughly two percent of your body weight, consumes about twenty percent of your total energy continuously, whether you are solving a problem or asleep — something on the order of 120 grams of glucose a day.

No glucose, no thought.

There is one honest nuance here. During prolonged fasting, the liver converts fat into ketone bodies, which the brain can use for a substantial share of its fuel — which is what makes extended fasting survivable. But even then, a portion of the brain's demand remains strictly glucose, and the body will manufacture that glucose out of its own tissue rather than let the level fall too far.

Which is the point. Your body defends its blood glucose tenaciously, from both directions, using multiple redundant hormonal systems. If you eat no carbohydrate at all, your liver will make what you need. The amount of glucose in your blood is a tightly regulated variable, not a passive readout of your last meal.

So the question is never whether there is glucose in your bloodstream. There is, and there must be. The question is what it takes to get it where it's going.

The arrival problem

Glucose has one inconvenient property, and it is the opposite of cholesterol's.

Cholesterol can't dissolve in water, so the hard part is travel — the body has to build carrier particles to move it through a water-based bloodstream. Glucose dissolves in water beautifully. Travel is trivial. Glucose reaches every cell in your body within minutes of a meal.

The hard part is the last few nanometers.

Every cell in your body is wrapped in a membrane made of lipid — a wall of fat, two molecules thick, that exists specifically to keep the inside of the cell separate from the outside. A water-soluble molecule cannot simply diffuse through a wall of fat. Glucose can arrive at the doorstep of every cell you own and still not get in.

So the body built a door, and then put a lock on the door, so that entry could be controlled rather than automatic.

The door. The door itself is a protein called a glucose transporter — a channel that spans the membrane and ferries glucose from outside to inside. There is a family of them, and which one a tissue uses tells you what that tissue is for:

• GLUT1 — everywhere, especially red blood cells and the blood-brain barrier. Always present on the surface. No key required.
• GLUT2 — liver and pancreatic beta cells. A low-affinity, high-capacity transporter that lets glucose flow in proportion to how much is in the blood, which is exactly what a sensor needs.
• GLUT3 — neurons. High affinity, so the brain wins the competition for glucose when supply is scarce.
• GLUT4 — skeletal muscle and fat tissue. The only one in the family that is insulin-responsive, and the one this entire story turns on.

The critical fact about GLUT4 is that it does not sit permanently on the cell surface. It waits inside the cell, packaged in small vesicles, and only moves up to the membrane when the cell is told to let glucose in. That movement is called translocation, and it is a physical, mechanical process — a hatch being carried into position and opened.

The key. The signal to open is insulin, a hormone made by the beta cells of your pancreas. Insulin binds the insulin receptor on the cell surface — the lock — and sets off a cascade inside the cell: the receptor phosphorylates a docking protein called IRS-1, which activates an enzyme called PI3K, which activates Akt, which finally releases the brake holding the GLUT4 vesicles in place.

The hatch translocates. It docks at the membrane. Glucose flows in. Blood glucose falls, insulin subsides, and the hatches are pulled back inside.

That chain of four or five steps is worth holding onto, because insulin resistance is not a failure of the key or the lock. It is a failure somewhere in the middle of that chain — and knowing that is what makes the disease legible.

The exception that will matter later. Your brain does not use this door. Neurons and the blood-brain barrier use GLUT1 and GLUT3, which sit permanently on the surface and require no insulin at all. Evolution was not going to make thinking contingent on a hormone.

But that privilege comes with a cost. A cell with a lock can regulate its intake. A cell without one takes what the blood is offering. When blood glucose is steady, the brain gets a steady supply. When blood glucose spikes and then plunges, the brain rides the spike and then rides the plunge — which is one reason the earliest perceptible symptom of metabolic drift is not a symptom of the blood at all. It is a symptom of cognition. Fog. Heaviness. The three-o'clock wall.

What a meal actually does

Most explanations stop at "you eat, insulin rises, glucose goes into cells." The real sequence is more interesting, and three of its details explain things the simple version cannot.

Insulin arrives in two phases. Your beta cells keep a small supply of insulin pre-made, sitting in granules, ready to go. When glucose rises, that stock is dumped almost immediately — a sharp spike within the first two to ten minutes, called first-phase insulin secretion. Only then does the pancreas begin manufacturing and releasing more, in a slower sustained wave: second phase.

First phase is the pre-emptive strike. Its job is to tell the liver to stop producing glucose before the meal's glucose has even finished being absorbed — to get out of the way of incoming supply.

The loss of first-phase insulin is one of the earliest measurable beta-cell defects in type 2 diabetes, and it appears long before fasting glucose moves. When it goes, the meal's glucose arrives while the liver is still adding its own, the peak is higher, and the pancreas has to chase it afterward with a larger, later second-phase release. That "late and large" pattern is the signature of an early problem, and it is invisible to every fasting test.

Your gut votes before your blood does. Swallow 75 grams of glucose and you release substantially more insulin than if the identical amount is infused directly into a vein to produce the same blood levels. The difference — often half or more of the total insulin response to a meal — comes from hormones released by the intestine as food passes through: GLP-1 and GIP, collectively the incretins.

Incretins amplify insulin release, suppress glucagon, slow the rate at which the stomach empties, and signal satiety to the brain. They are the reason a meal is handled better than the sugar in it would predict. Blunted incretin signaling is part of the type 2 picture — and the GLP-1 receptor agonists now used as medications are, in essence, a pharmacological amplification of this native system.

There is a second hormone, pointing the other way. The pancreas also makes glucagon, from alpha cells, and glucagon does the opposite of insulin: it tells the liver to release glucose. Your blood glucose at any moment is set less by insulin alone than by the ratio of insulin to glucagon.

After a meal, glucagon should fall. In type 2 diabetes it often doesn't — it stays inappropriately high, so the liver keeps adding glucose to a bloodstream that is already flooded. This is one of the least-discussed contributors to post-meal spikes, and one of the reasons the post-meal window is where the disease reveals itself first.

Where does the glucose actually go? Overwhelmingly, into muscle. Skeletal muscle accounts for roughly eighty percent of insulin-stimulated glucose disposal. Muscle is not a bystander in metabolic health; it is the primary destination and the largest reservoir.

Once inside, glucose is either burned or stored as glycogen — glucose molecules linked into a dense branched chain. Your liver holds roughly 100 grams; your skeletal muscle holds roughly 400 grams, several times more.

And the two stores are not interchangeable. Liver glycogen can be broken down and exported back into the bloodstream to supply the rest of the body. Muscle glycogen cannot leave the muscle — it lacks the enzyme required to release free glucose — so it is a private reserve, filled by that muscle and spent by that muscle.

Which has a consequence people rarely connect. If muscle glycogen is chronically full, because the muscle is rarely emptied by work, then a large share of the body's glucose sink is unavailable. Exercise doesn't just burn glucose. It creates somewhere for the next meal to go.

The overnight shift

There is a second source of blood glucose, and almost nobody thinks about it — which is unfortunate, because it is the source the standard blood test actually measures.

You do not eat for eight to twelve hours every night. Your brain does not stop consuming glucose during that time. Your red blood cells never stop. So something has to supply it.

That something is your liver.

Early in the night, the liver simply unpacks its glycogen and releases the glucose — glycogenolysis. As those stores draw down over the course of the night, a second mechanism takes over progressively: the liver manufactures new glucose from other raw materials. This is gluconeogenesis, literally "the making of new glucose," and by morning it is supplying much of your circulating glucose.

The raw materials are worth being precise about, because this is commonly mangled:

• Lactate, recycled from muscle and red blood cells (the Cori cycle).
• Amino acids, principally alanine, drawn from protein.
• Glycerol, the three-carbon backbone released when a stored triglyceride is broken apart.

That last one is the link to fat. Lipolysis — the release of stored fat from adipose tissue — sends fatty acids and glycerol into circulation. The glycerol can become glucose. The fatty acids themselves cannot be converted into glucose in humans; they are burned directly as fuel or converted into ketones. It is a small but important distinction: fat mostly fuels the process of gluconeogenesis rather than becoming the glucose itself.

This whole system has a thermostat, and the thermostat is insulin. Low insulin overnight permits fat release and hepatic glucose production. Rising insulin in the morning says: stop, supply has arrived.

There is also a scheduled disturbance. In the pre-dawn hours, cortisol and growth hormone rise as part of the normal circadian pattern, and both push the liver to produce more glucose — priming you to wake up. This is the dawn phenomenon. In a healthy system it is invisible. In an insulin-resistant one, it is often why the morning reading is the highest of the entire day.

Two conclusions follow, and both matter enormously.

First: your fasting glucose is a liver number. It is not a verdict on last night's dinner. It is a report on how well your liver's overnight production is being regulated. When fasting glucose rises, the story is usually not "too much sugar last night." It is the liver is no longer standing down when it's told to.

Second: insulin is the off-switch for fat release. High insulin doesn't only open the glucose hatches; it simultaneously shuts down lipolysis. That's exactly what you'd want after a meal — no sense unpacking stored fuel while fuel is arriving. But it means a body running chronically high insulin is a body that has partially locked its own fat stores.

That is a trap, and we'll come back to it.

The tank, and the spillover

Now: where does the trouble actually begin?

Not in the blood. In storage.

Fat is not a design flaw. It is one of the most elegant solutions in biology — nine calories per gram, chemically stable, storable for months, the reason humans could survive seasons of scarcity and afford the metabolic luxury of a large brain. The body has a compartment purpose-built to hold it: subcutaneous adipose tissue, the fat under your skin.

Think of it as a tank. As long as excess energy goes into the tank, the system is doing its job. Subcutaneous fat is metabolically fairly quiet, and it is capable of expanding — both by enlarging existing fat cells and by recruiting new ones. A person can carry a great deal of it and remain metabolically healthy.

But the tank has a capacity, and here is the part that surprises almost everyone: that capacity is personal, and substantially set by genetics. Two people can carry identical amounts of extra weight and be in completely different metabolic situations, because one has a large, expandable, well-behaved tank and the other has a small one that filled years ago. This is the personal fat threshold.

The most striking evidence for it comes from the extremes. People with lipodystrophy — a condition in which subcutaneous fat tissue is largely absent — develop severe insulin resistance, fatty liver, and diabetes despite being visibly very lean. They have almost no tank at all, so everything spills immediately. This is close to a natural experiment, and it demonstrates that the problem is not the quantity of fat in the body but whether there is a safe place to put it.

The same logic explains the opposite case: people carrying substantial subcutaneous fat with entirely normal insulin sensitivity, sometimes for decades. Their tank is large and it is holding.

It also explains one of the most consequential observations in global health. South Asian populations develop insulin resistance, fatty liver, and type 2 diabetes at body weights that Western risk calculators consider unremarkable — which is why diagnostic BMI thresholds are set lower for them. Less capacity to store fat safely, at the same visible size.

When energy keeps arriving after the tank is full, it doesn't stop arriving. It spills into compartments never designed to hold much of it:

• Around the organs — visceral fat, packed into the abdominal cavity, metabolically noisy, inflammatory, and draining directly into the portal vein that feeds the liver.
• Into the liver — hepatic steatosis, "fatty liver," now the most common liver condition in the developed world.
• Into the pancreas — fat infiltrating the very organ that makes insulin.
• Into muscle cells themselves — intramyocellular lipid: fat droplets sitting inside the cell that is supposed to be taking up glucose.

That last one is where the disease lives.

Fat in the wrong compartment is called ectopic fat, and ectopic fat is not inert storage. It is a chemical problem in a place with no way to handle it.

The jam

Here is the mechanism, in the plainest terms available.

When fat accumulates inside a muscle cell faster than the cell can burn it, the intermediates of fat metabolism build up — particularly diacylglycerol (DAG) and ceramides. These are not inert. DAG activates an enzyme (PKCθ) that chemically modifies IRS-1, the docking protein in the insulin signaling chain, in a way that stops it working properly. Ceramides interfere further downstream, blocking Akt.

Recall the chain: receptor → IRS-1 → PI3K → Akt → GLUT4 translocation. The fat doesn't touch the key or the lock. It breaks the wire between the lock and the hatch.

So insulin binds normally, the receptor fires normally, and the hatch doesn't come up. Glucose clears from the blood more slowly than it should.

Your pancreas has one lever, and it pulls it. It makes more insulin. And it works — with enough signal pushed through a degraded wire, enough hatches eventually open and the glucose clears. Blood glucose returns to normal.

This is the central fact of the entire disease. For well over a decade, blood glucose stays normal not because the system is healthy but because the system is compensating. The normal result is being manufactured at a rising cost, and that cost is invisible to every test that measures the result.

The term insulin resistance is usually explained as "the cells stop responding to insulin," which makes it sound like a behavior, or an attitude. It is closer to the truth to say the wiring is fouled with fat, and the pancreas is shouting to be heard.

And then there is the detail that ties the whole picture together.

Insulin resistance is selective. Insulin gives the liver two separate instructions: stop producing glucose, and convert surplus into fat for storage (de novo lipogenesis). In the insulin-resistant liver, the first instruction stops being obeyed while the second one keeps working — arguably works harder, because insulin levels are now so high.

That is why the numbers on a metabolic panel look the way they do. The liver is simultaneously ignoring insulin (glucose output stays high) and over-obeying insulin (fat production stays high). High glucose and high triglycerides, in the same person, at the same time, from the same hormone.

It is not a contradiction. It is one broken wire and one intact one.

The shouting

The compensation keeps glucose normal for years. But the compensation is not free, and chronically elevated insulin — hyperinsulinemia — does damage of its own, in several directions at once.

It locks the fat in. Insulin suppresses lipolysis. So the state caused by too much stored fat is a state in which stored fat becomes harder to release. The disease defends itself, and this is a large part of why "just lose weight" is such poor advice delivered without a mechanism: the hormonal environment is actively working against the instruction.

It drives the liver to make fat. Through de novo lipogenesis, high insulin converts surplus substrate into triglycerides. Some are exported into the blood; some stay behind, worsening the fatty liver that caused the resistance in the first place. A loop that feeds itself.

It rewrites your lipid panel — years before glucose moves. Rising triglycerides and falling HDL are among the earliest visible fingerprints of insulin resistance. The same process shifts LDL particles toward the smaller, denser, more atherogenic pattern and pushes ApoB particle count up.

This is worth stating plainly: the metabolic disease and the cardiovascular disease are the same disease, read off two different lab panels. Cardiovascular risk in people who eventually develop type 2 diabetes begins rising during the prediabetic years — well before the glucose threshold is crossed. The heart attack is not a late complication of diabetes. Both are downstream of the same drift.

It raises blood pressure. Insulin promotes sodium retention in the kidney and affects sympathetic tone. This is one reason blood pressure so often travels with insulin resistance, and why the cluster acquired a name — the constellation of central adiposity, high triglycerides, low HDL, elevated blood pressure, and elevated fasting glucose that clinicians recognize together.

And it wears out the pancreas. Beta cells are not an infinite resource. Two decades of maximal output produces measurable decline in capacity — which is the subject of section XI, and the reason the timing of intervention matters more than almost anything else in this disease.

The exhaust

There is one more marker in this system, and it sits at the opposite end from the ones we've been discussing.

Glucose and insulin are upstream. They are the inputs and the effort — what is arriving, and what the body is spending to handle it.

Uric acid is downstream. It is a waste product: the end point of purine metabolism, produced when the cell recycles the building blocks of DNA, RNA and ATP, and cleared through the kidneys. It is what comes out the back of the system after everything else has run.

Two mechanisms push it up, and a strained metabolism produces both.

The liver makes more of it. This is where fructose is genuinely different from glucose. Glucose entering a cell is phosphorylated by an enzyme that is regulated by feedback — when the cell has enough, the process slows. Fructose is phosphorylated by fructokinase, which has no such brake. A large fructose load is therefore taken up and phosphorylated rapidly and without restraint, transiently depleting the cell's ATP. The spent ATP degrades to AMP, and AMP is degraded onward to uric acid. So a liver repeatedly handling large fructose loads is a liver repeatedly generating uric acid — while, in the same process, running de novo lipogenesis and contributing to fatty liver.

The kidney holds on to more of it. Insulin stimulates urate reabsorption in the renal tubule. So the higher your circulating insulin, the less uric acid you excrete. Hyperinsulinemia doesn't just create more; it clears less.

Which is why the reading is so informative, and so routinely ignored.

A high uric acid on an otherwise unremarkable panel is not a separate finding that happened to land on the same page. It suggests the whole line has been running hot for a long time — intake, hepatic handling, storage, hormonal signaling, and renal clearance, all under strain simultaneously. Upstream markers tell you what is happening now. A downstream marker tells you the system has been accumulating strain end to end, for years.

Its most famous consequence is gout — urate crystallizing in a joint. But metabolically, the more useful reading is as an integrated signal of how long the system has been under load. There is also active research into whether elevated uric acid contributes to endothelial dysfunction and hypertension directly, rather than merely marking them; that question is not settled, and it is worth holding loosely.

What is not in dispute is the association. Uric acid tracks with visceral fat, fatty liver, triglycerides, blood pressure, and insulin resistance — and it is sitting on lab panels, out of range, being ignored, in an enormous number of people who are told their results are fine.

The four dominoes

Now the whole twenty years can be laid out. Once you see the order, the diagnostic blind spot becomes almost absurd.

Years 0–5: post-meal insulin rises. The first measurable abnormality. After a meal, the pancreas must release substantially more insulin than it once did to clear the same glucose load. Clearance still happens; it just costs more. Somewhere in here, first-phase insulin secretion begins to blunt, so the peak runs higher and the correction arrives later. Every glucose measurement — fasting, post-meal, A1c — is normal. The person feels normal. A standard annual panel: perfect.

Years 5–10: fasting insulin rises. The compensation is no longer confined to meals. It now takes elevated insulin around the clock, including all night, simply to hold the morning glucose where it has always been. This is where the disease becomes systemic — and, crucially, where it becomes detectable by a single inexpensive blood test drawn at the exact hour blood is already being drawn.

It is also where the other numbers start moving. Triglycerides climb. HDL falls. ApoB drifts up. Uric acid rises. Liver fat accumulates, and ALT may nudge upward. Blood pressure creeps. Fasting glucose may go from 88 to the mid-90s — a change nobody looks at twice. Standard annual panel: essentially unremarkable, with a handful of numbers "slightly off" that no one connects to each other.

Years 10–15: post-meal glucose breaks. The pancreas can still win overnight but can no longer win after a meal. Glucose overshoots after eating; the insulin response arrives late and excessive; glucose then falls fast, and further than it should. High insulin simultaneously blocks fat release, so the usual backup fuel is unavailable during the descent. The brain, whose transporters have no lock and no thermostat, experiences the whole ride.

This is where symptoms finally appear — and they are not the symptoms anyone is screening for. Post-meal fatigue. Afternoon fog. Cravings that arrive on a schedule. Hunger ninety minutes after eating. These get coded as stress, poor sleep, or age.

Fasting glucose now reads roughly 100–110. A1c drifts to 5.8–6.2. This is the "prediabetes" window, which in standard practice most often produces the advice to watch your diet and come back in a year.

Years 15–20: fasting glucose breaks. The liver, fatty and no longer regulated, keeps pouring glucose into the blood overnight, amplified each dawn by cortisol and growth hormone. Glucagon fails to fall appropriately after meals. And the pancreas, after nearly two decades at maximum, is losing capacity. For the first time, an entire night of maximal insulin cannot restore the morning number.

Fasting glucose crosses 126. A1c crosses 6.5. The threshold is met, the diagnosis is made, the prescription is written.

Four dominoes, in a consistent order: post-meal insulin, fasting insulin, post-meal glucose, fasting glucose. The annual panel watches the fourth.

The pace varies between people — some move through this in twelve years, some in thirty, and some stall indefinitely at a stage. But the sequence is remarkably conserved, and so is the blind spot. The standard screening architecture is structurally incapable of detecting this disease during the fifteen or so years in which it is most reversible. Not because the test is wrong, but because the test measures the outcome the body is working hardest to protect — and notices failure only after the compensation has been exhausted.

The beta cell, and why timing decides everything

Everything above describes a system under load. This section describes the part that eventually breaks, because it is the part that determines whether any of it can be undone.

Your beta cells are not simply insulin pumps. They are glucose sensors that respond to a rise in blood glucose by secreting a proportionate amount of hormone — and for years of compensation they do something remarkable: they expand. Beta-cell mass increases, output per cell increases, and the system successfully covers a growing shortfall.

Then it stops covering it.

The classic UKPDS data suggested that by the time of diagnosis, roughly half of beta-cell function is already lost, and that function continues declining afterward regardless of which standard therapy is used. That is the most damning single statistic in this disease. It means diagnosis occurs not at the beginning, and not even at the middle, but well into the failure phase.

What is lost turns out to matter enormously. Two mechanisms are in play:

Apoptosis — beta cells dying and not being replaced. This is not recoverable.

Dedifferentiation — beta cells that are still alive but have reverted toward a more primitive, non-functional state, essentially switching off their identity as insulin producers under chronic metabolic stress. This is potentially recoverable, and evidence suggests it accounts for a substantial share of the lost function.

That second mechanism is the biological basis for remission. Cells that switched off can, in the right conditions, switch back on. Cells that died cannot. And the balance between the two shifts steadily toward the irreversible with duration of disease.

Two further stressors compound this. Glucotoxicity: chronically elevated glucose is itself toxic to beta cells, so once glucose starts rising, the rise accelerates its own cause. Lipotoxicity: fat infiltrating the pancreas impairs beta-cell function directly, which is why removing fat from the pancreas specifically — not simply losing weight generally — appears to be central to recovery.

This is why the same intervention produces such different results at year eight and year twenty-two. It is not motivation and it is not adherence. It is how much of the loss is dedifferentiation and how much is death.

The disease is most reversible in exactly the window where nothing is being measured.

What A1c actually measures — and what it hides

Hemoglobin A1c has become the defining number of this disease, so it deserves to be understood rather than trusted.

Glucose in your blood spontaneously and irreversibly attaches to proteins it encounters — a slow, non-enzymatic process called glycation. Hemoglobin, sitting inside red blood cells that circulate for about three to four months, accumulates this modification in proportion to the glucose it has been bathed in. Measure the fraction of glycated hemoglobin and you get an integrated estimate of average glucose over the lifespan of those cells.

That is genuinely useful. It is also three things people rarely appreciate.

It is weighted toward the recent past. Roughly half of an A1c value reflects the preceding month, with the earlier months contributing progressively less. It is not a clean three-month average.

It depends on red blood cells behaving normally. Anything that changes red cell lifespan changes A1c independent of glucose:

• Anemia from blood loss, hemolysis, or high red cell turnover → cells are younger on average → falsely low A1c.
• Iron-deficiency anemia → cells older on average → falsely high A1c.
• Chronic kidney disease, recent transfusion, pregnancy, and certain hemoglobin variants → unreliable in either direction.

Even between healthy people, average red cell lifespan varies enough to matter. Two people with identical true average glucose can produce meaningfully different A1c values — a discrepancy sometimes called the glycation gap.

And most importantly: it is a mean, and a mean cannot see variability. A person whose glucose sits placidly at 120 all day and a person who swings between 70 and 200 after every meal can produce the same A1c. Metabolically, they are not in the same situation — variability itself is associated with worse outcomes, and it is the swinging person who is having the afternoons.

This is precisely what a continuous glucose monitor recovers and a single A1c discards. A1c is a good longitudinal marker of overall glycemic burden and a poor description of what is actually happening to someone across a day.

The measurement menu

Here is what can actually be measured, what each thing answers, and what each one's limits are.

Fasting insulin. The earliest practical marker. Same fast, same morning, same draw, one more tube; low cost; on every major lab menu. It answers: how much effort is this system spending at rest? A fasting glucose of 92 produced with a fasting insulin of 4 is a healthy body. The same 92 produced with a fasting insulin of 25 is a body that has been compensating for years.

Its honest limits: insulin assays are not well standardized between labs, so absolute values are not perfectly comparable and trends within one lab are more meaningful than single values across labs. And the "reference range" printed on the report is population-derived — meaning it was built from a population in which insulin resistance is extremely common, so being "within range" is a weak reassurance.

HOMA-IR. Fasting insulin and fasting glucose combined into a single index: (fasting insulin in µIU/mL × fasting glucose in mg/dL) ÷ 405. It answers the same question with a bit more resolution and travels better between contexts. It is an estimate of hepatic insulin resistance specifically, since fasting glucose is a liver number.

Oral glucose tolerance test, with insulin measured alongside glucose. The deep version: a glucose load, then paired glucose and insulin sampled over two to three hours. This is the only common test that can see domino one directly — the shape and timing of the insulin response, including the loss of first phase. It is rarely ordered with insulin included, which is a shame, because the glucose-only version discards most of the information.

Continuous glucose monitoring. Two weeks of sensor wear, one insertion. It answers: what actually happens after meals, across afternoons, and overnight? The useful readouts are not just the average but the post-meal peak and two-hour value, the variability (standard deviation, or the amplitude of the swings), the overnight baseline, and time in range.

Its honest limits: interstitial fluid lags blood by several minutes, accuracy is imperfect at extremes, and it is entirely possible to over-interpret a single spike. Its value is in pattern recognition across two weeks, not in judging one lunch.

Triglyceride-to-HDL ratio. A free proxy hiding on a lipid panel you probably already have; a higher ratio correlates with insulin resistance and with the small-dense-LDL pattern. Its limit: it performs notably less well in Black patients, in whom triglycerides tend to run lower at any given level of insulin resistance — so a reassuring ratio is not reassuring in everyone.

Liver fat. ALT is a crude and insensitive hint. Ultrasound is widely available. FibroScan with CAP, or MRI-PDFF, actually quantify it. Fatty liver is often the first structural evidence of spillover and one of the most responsive to intervention.

Body composition and fat distribution. Waist-to-height ratio (keep it under half) is free and outperforms BMI. Visceral fat estimates, DEXA, and MRI add resolution. The question is never "how much do you weigh" — it is "where is it, and have you exceeded your own threshold."

Uric acid. Almost always already on the panel. Read as described in section IX: a downstream integrator, not a curiosity.

ApoB. Because the cardiovascular disease is the same disease. Discussed in full in the atherosclerosis canon.

What actually changes the trajectory

The mechanism above dictates the interventions. This is not a lifestyle list; each item maps to a specific step in the biology.

Contraction opens the hatch without the key. This is the single most important practical fact in the entire canon. Muscle contraction triggers GLUT4 translocation through a pathway that does not depend on the insulin receptor at all — signaling via AMPK and calcium, converging on the same machinery from a different direction.

Two routes to the same hatch. When the insulin route is fouled with fat, the contraction route still works.

This is why ten to fifteen minutes of walking after a meal measurably blunts the post-meal excursion, and why the effect is largest in exactly the people whose insulin route is most impaired. It is not a metaphor for discipline. It is a parallel biochemical pathway that partially bypasses the broken one. And the effect of a single bout of exercise on insulin sensitivity persists for roughly twenty-four to forty-eight hours — which makes frequency more valuable than intensity, and makes "never two days off" a more useful rule than any single workout.

Muscle is the sink, so build the sink. Skeletal muscle handles the large majority of insulin-mediated glucose disposal, and muscle glycogen is a private store that only that muscle can fill or spend. Resistance training raises capacity rather than merely lowering load — one of the few interventions in medicine that does. It is also the intervention most neglected in people over forty, precisely when it starts to matter most.

Empty the tank's overflow, not just the tank. The specific goal is removing fat from the liver and pancreas, which is why remission tracks with the magnitude of fat loss and why liver fat responds early — often within weeks, well before the scale reflects much. This is also why the target is personal: the amount required is however much brings you back under your threshold, which may be a great deal less, or considerably more, than population averages suggest.

Meal composition and order. Protein, fat, and fiber consumed before or alongside carbohydrate reduce the post-meal glucose excursion, partly by slowing gastric emptying and partly through incretin signaling. The effect is real and modest; it is best used as one lever among several, and best verified against your own CGM rather than assumed.

Sleep is not adjacent to this — it is inside it. Experimental sleep restriction in healthy young adults reduces insulin sensitivity measurably within days; the classic studies produced impairments in glucose tolerance after less than a week of short nights. Sleep loss also raises evening cortisol and shifts appetite signaling. A person fighting their post-meal glucose while sleeping five hours is fighting the biology from both ends.

Remission is real, and it is time-dependent. Structured intensive weight-management trials have produced diabetes remission — normal glucose off medication — in a substantial minority to near-half of participants at one year, with a meaningful share sustaining it at two years. Remission correlated strongly with the amount of weight lost and, critically, with shorter duration of diabetes.

The honest framing: remission is well documented, far more achievable early, considerably less so after long duration, not universal, and not the same thing as cure — the underlying tendency remains, and regain generally means relapse. It is also a decision that belongs with a physician, particularly where medications that can cause hypoglycemia are involved.

But "you will manage this for life" describes the average outcome of standard care. It does not describe the biology.

Why medicine measures it this way

If the disease is this legible, why is the screening this narrow? The answer is historical, and understanding it makes the whole system less baffling.

The thresholds were set for a different purpose than early detection. The diagnostic fasting glucose cutoff of 126 was chosen because it is roughly the level above which the risk of retinopathy — visible damage to the small vessels of the eye — begins to climb steeply. The A1c cutoff of 6.5 was adopted later on similar reasoning, with the added advantage that it requires no fasting.

Read that again. The thresholds mark the point at which complications become likely, not the point at which the disease begins. They were designed to identify people who need treatment now, and they do that job reasonably well. They were never designed to identify a metabolic trajectory fifteen years out, and they cannot.

Insulin was never adopted as a screening test. There is no standardized assay, no agreed reference range for "optimal" versus merely "average," and no clinical trial demonstrating that treating an isolated high fasting insulin improves hard outcomes — largely because that trial has not been run. Guidelines are built on that kind of evidence, so the test sits outside them.

And there is no drug indicated for it. Screening in practice tends to concentrate around numbers with a treatment attached. A number that would mainly prompt a conversation about sleep, muscle, and food has weak institutional gravity, no matter how early it moves.

None of this is conspiracy. It is the ordinary consequence of a system optimized to treat disease rather than to observe trajectories — one built around thresholds, visits, and interventions rather than slopes.

But it does mean that "your labs are normal" is a statement about a threshold, not a statement about your direction. Those are very different claims, and only one of them was ever tested.

Why this matters for how you read your own report

If you have a recent panel, look at it now.

It probably has a fasting glucose. It may have a hemoglobin A1c. If you're lucky it has triglycerides, HDL, and a uric acid. It almost certainly does not have a fasting insulin. It contains no information about what happens to you after a meal. It contains no measure of where your fat is stored. It contains nothing about the effort required to produce the numbers it does show.

A physician reading that report has been trained to look at one number. Under 100, the conversation is see you next year. Between 100 and 125, it is usually watch your diet, see you next year. At 126, it becomes a prescription.

But you now know what the report doesn't show.

You know the disease falls in a consistent order across roughly twenty years, and that the number on your page is the last of the four to move. You know a normal glucose can be a compensated glucose, and that nothing on the page distinguishes the two. You know that triglycerides and HDL drifting in opposite directions is an early signature, not an unrelated finding. You know that a high uric acid means the whole line has been running hot. You know that an A1c is a mean, and a mean cannot see your afternoons. You know that by the time the threshold is crossed, half the beta-cell function is typically gone — and that a meaningful share of what's lost may be recoverable if it is caught while it is still switched off rather than dead.

And you know the mechanism well enough to know what actually changes it: getting fat out of the compartments never built to hold it, and using the second route to the hatch — the one that runs through your legs — while you do.

The threshold is a moment. The slope is the truth. And the slope has been in your own chart, in your own numbers, the entire time.