Skip to content
PeptideHormone

The fat hormone that falls as fat rises

Adiponectin is made almost entirely by fat cells and fills the blood at levels far above most hormones, yet people with more fat have less of it. Low adiponectin predicts type 2 diabetes about as reliably as any blood marker, and in mice adding it reverses insulin resistance. So why is there no adiponectin drug? Partly because the protein is too large and complex to replace, and partly because human genetics suggests it is a readout of unhealthy fat more than a cause of disease. The two adiponectin paradoxes, and what a hormone that never became a medicine has to teach.

8 min read · reviewed November 2026

Found four times in two years

In the mid-1990s, several laboratories went looking for what fat cells make, and kept finding the same thing. One group cloned a protein secreted only by fat cells and called it Acrp30 (Scherer et al., 1995). Another found a fat-specific gene and called it AdipoQ (Hu et al., 1996). A third sequenced the messages in human fat tissue and found that one outnumbered all the others, so they named it apM1, for adipose most abundant gene transcript (Maeda et al., 1996). A fourth pulled the protein out of blood plasma. It was one molecule, and it eventually got one name: adiponectin.

Two things about it stood out immediately. Fat tissue was pouring effort into making it, and the blood was full of it. Adiponectin circulates at concentrations of micrograms per millilitre, far above the level of nearly every other hormone in this catalog. Whatever it was for, the body treated it as important. Nobody yet knew what it did.

The first paradox

The obvious prediction was simple. A protein made only by fat should be higher in people with more fat. That is how leptin behaves, the fat hormone discovered a year earlier: more fat, more leptin, in a nearly straight line.

When a reliable blood test for adiponectin was finally built, it showed the reverse. People with obesity had less adiponectin than lean people, a result the authors put in the title of their paper as a “paradoxical decrease” (Arita et al., 1999). More of the tissue, less of its product.

The low level turned out to mean something. Among Pima adults matched for body size, those with lower adiponectin were more likely to develop type 2 diabetes later (Lindsay et al., Lancet, 2002). Across thirteen prospective studies and more than fourteen thousand people, the pattern held in every ethnic group examined: the higher the adiponectin, the lower the risk of diabetes (Li et al., JAMA, 2009). Few blood markers track metabolic health as consistently.

Leptin and adiponectin compared: both are made by fat cells, but they respond in opposite directions as fat mass rises and report different things.
PropertyLeptinAdiponectin
Made byFat cellsFat cells
As fat mass risesRisesFalls
What it reportsHow much fat is storedHow well the fat tissue is working
Level in bloodLow, like most hormonesVery high for a hormone
Human geneticsLosing the gene causes severe obesityLow-adiponectin variants do not raise diabetes risk
As a drugApproved for rare deficiency statesNever tested as a therapy in people
Two hormones from the same cell, moving in opposite directions. Leptin measures the size of the fat store; adiponectin tracks its condition.

The case that it protects

In mice, adiponectin looked like much more than a marker. Giving it to obese, insulin-resistant animals lowered the fat content of their liver and muscle and restored their response to insulin (Yamauchi et al., 2001), and it made the liver more sensitive to insulin’s signal to stop releasing glucose (Berg et al., 2001). Two years later its receptors were cloned, AdipoR1 and AdipoR2, giving the hormone a proper signaling pathway (Yamauchi et al., Nature, 2003).

The most striking experiment took mice that lack leptin, which eat without limit and become obese and diabetic, and engineered them to make a little extra adiponectin. The animals got even fatter. They were also metabolically healthy: normal glucose, normal insulin, less inflammation in their fat (Kim et al., 2007). The extra fat had gone into fat tissue, where it is stored safely, and stayed out of the liver and muscle, where it does damage.

The idea this suggests

The harm in obesity may come less from how much fat there is than from where it ends up. On this reading adiponectin is the signal that keeps fat tissue expanding and accepting storage, so the overflow does not land in other organs.

It also explains the first paradox. Adiponectin output depends on the state of fat cells, not their count. Small, healthy, insulin-responsive fat cells secrete it freely. Overfilled, inflamed ones make less (Wang & Scherer, 2016). Leptin reports how much fat is stored. Adiponectin reports how well the storage is working.

Marker or cause?

So low adiponectin predicts diabetes in people, and adding adiponectin fixes insulin resistance in mice. It is tempting to join those two facts and conclude that low adiponectin causes diabetes in people. There is a way to test that, and the test did not support it.

Some people inherit variants of the adiponectin gene that give them lower levels for life. If adiponectin protects against diabetes, those people should be more insulin resistant and get more diabetes. A study pooling genetic data from tens of thousands of people looked for exactly that. In ordinary observational analysis, lower adiponectin went with worse insulin sensitivity, as expected. But the genetic variants that lower adiponectin did not carry the predicted increase in insulin resistance or diabetes. The authors concluded their data did not support a causal role (Yaghootkar et al., 2013).

That leaves the simplest reading of the human data: unhealthy fat tissue causes both things, low adiponectin and diabetes, and the two travel together without one producing the other. The mouse experiments are real, but they add far more adiponectin than a gene variant takes away, and a mouse is not a person. This is the gap described in mechanism is not proof, in an unusually clean form.

The drug evidence has the same ambiguity. The thiazolidinediones, a class of diabetes drugs that act on fat cells, raise adiponectin sharply. In one study three months of treatment roughly doubled it, alongside improved insulin sensitivity (Yu et al., 2002). But those drugs remodel fat tissue in many ways at once. The rise in adiponectin may be how they work or may just be a sign that the fat is healthier.

The second paradox

If high adiponectin marks good metabolic health, people with high levels should live longer. In some groups the opposite is seen. Among 195 patients with chronic heart failure, those with higher adiponectin were about three times as likely to die over the following years as those in the lowest third (Kistorp et al., 2005). A later analysis pooling 55 studies found the same direction across the board: each step up in adiponectin went with a higher death rate (Scarale et al., 2018).

The explanation again seems to be that adiponectin is reporting on something else. Illness and wasting raise it. So do the natriuretic peptides that a failing heart releases, and when the pooled analysis accounted for those, a good part of the association went away. A sick, thin, older person has high adiponectin for the same reason a lean, healthy young one does, which is that there is little overfilled fat. The number is the same and the meaning is different.

Reading the number

Adiponectin is a faithful marker with a context problem. High is good news in a healthy forty-year-old and a warning sign in an eighty-year-old with heart failure. A blood level on its own does not say which.

Why there is no adiponectin drug

Even setting the causality question aside, adiponectin is a hard molecule to turn into a medicine, for reasons that say something about the limits of the peptide playbook.

Most hormones in this catalog are short chains that a machine can build. Adiponectin is a full-sized protein that cells chemically modify and then assemble into groups of three, six, or a dozen and more, and the largest assemblies appear to be the most active (Wang & Scherer, 2016). That assembly needs living mammalian cells and is hard to control. And because the natural level in blood is so high, a replacement dose would be enormous compared with any hormone given today. It sits at the far end of the complexity ladder.

So the field went around the protein and aimed at its receptors. A small molecule taken by mouth, AdipoRon, activated both adiponectin receptors and improved insulin resistance in obese mice (Okada-Iwabu et al., Nature, 2013). A decade later the same group reported an antibody that switches the receptors on and might be dosed monthly (Asahara et al., 2023). Both results are in mice. No adiponectin-based therapy has been shown to help people, and given the genetic evidence, it is an open question whether one would.

What the hormone is telling us

Adiponectin is one of the best-studied hormones never to become a drug. Its story is a useful counterweight to the rest of this catalog, where the usual arc runs from mechanism to molecule to trial.

  • Made by fat, lower with more fat. Adiponectin comes almost entirely from fat cells and circulates at very high levels, yet it falls as fat mass rises. Its output tracks the health of fat tissue, not its amount.
  • A strong predictor. Low adiponectin precedes type 2 diabetes across populations. In mice, adding it reverses insulin resistance and steers fat into safe storage.
  • Probably not the cause. People who inherit lower adiponectin are not more prone to diabetes. In humans it looks more like a readout of unhealthy fat than a driver of disease.
  • High is not always good. In heart failure and old age, higher adiponectin goes with higher mortality, because illness and wasting raise it too.
  • No drug, for two reasons. The protein is too large, complex and abundant to replace, and receptor activators have only been shown to work in mice.

Educational reference on mechanism, summarized from public scientific literature and clinical-trial disclosures and simplified in places. Not individualized medical advice or a prescription. Where a dose or regimen appears it is reported from its source as research literacy — so you can read the evidence yourself — not an instruction to dose. Specific compounds and trials are named to explain the science; verify any claim against the linked primary sources.