What Is Adrenomedullin? The Peptide Named After the Wrong Organ

Peptide reference

Adrenomedullin is a 52-amino-acid vasodilator peptide that was named after the organ it was found in rather than the tissue that makes most of it — and that misnaming turns out to be the least surprising thing about it. Isolated in 1993 from a human pheochromocytoma, adrenomedullin has since become one of the better-validated prognostic markers in sepsis and acute heart failure, and the target of a drug programme built on a deliberately counterintuitive idea: an antibody designed not to block its target.

What is adrenomedullin?

Adrenomedullin was discovered by Kitamura and colleagues at Miyazaki Medical College, who screened extracts of human pheochromocytoma tissue for anything that raised platelet cyclic AMP. The peptide they pulled out was, in their own words, “a novel hypotensive peptide,” 52 amino acids long, carrying one intramolecular disulfide bond and showing “slight homology with calcitonin gene related peptide.” Because the material was abundant in normal adrenal medulla as well as in the tumour, they called it adrenomedullin.

That name has been quietly misleading ever since. Within a year, the same Japanese group reported that cultured endothelial cells expressed adrenomedullin messenger RNA at an intensity 20 to 40 times higher than adrenal gland, and secreted the peptide at a rate comparable to endothelin-1. Adrenomedullin is, functionally, an endothelial peptide. The adrenal gland was simply the tissue where a pheochromocytoma made it concentrated enough to purify in 1993.

Structurally it belongs to the calcitonin/CGRP superfamily, alongside calcitonin, CGRP and amylin. Like its relatives, it needs a C-terminal amide and a small disulfide ring to work at all: UniProt annotates a tyrosine amide at residue 146 and a disulfide between Cys110 and Cys115 of the precursor. A separate gene, ADM2 on chromosome 22, encodes intermedin (adrenomedullin 2) — a genuinely different peptide, not a fragment of this one.

One precursor, four peptides — and why the blood test is not the hormone

The ADM gene sits at 11p15.4 and encodes a 185-residue precursor. Processing yields a signal peptide, a second bioactive peptide called PAMP (proadrenomedullin N-terminal 20 peptide, residues 22–41), mature adrenomedullin at residues 95–146, and a C-terminal fragment. This matters clinically for a reason that catches people out.

Mature adrenomedullin is short-lived in plasma and circulates partly bound to a binding protein, which makes it awkward to assay reproducibly. In 2004 a group at what was then Brahms AG solved the problem sideways: they identified a fragment of the same precursor, proADM 45–92, that is “produced in stoichiometric amounts to Adrenomedullin and PAMP, but — contrary to them — is apparently non-functional and stable.” That fragment, mid-regional pro-adrenomedullin or MR-proADM, is what most published sepsis and heart-failure studies measure.

The receptor is defined by its chaperone, not by its gene

Adrenomedullin does not have a receptor of its own in the way most hormones do. It signals through a single G-protein-coupled receptor, calcitonin receptor-like receptor (CLR, gene CALCRL), which binds nothing on its own. What CLR responds to is decided by which receptor activity-modifying protein — RAMP1, RAMP2 or RAMP3 — is escorting it to the cell surface.

CLR with RAMP2 is the AM₁ receptor; CLR with RAMP3 is AM₂; CLR with RAMP1 is the CGRP receptor. The same three accessory proteins applied to the calcitonin receptor instead produce the three amylin receptors. Two GPCR genes, three chaperones, six pharmacologically distinct receptors.

The practical consequence is that adrenomedullin is not a selective ligand. Its reported potency at the CGRP receptor (pKi 8.1) sits roughly an order of magnitude below its potency at AM₁ (pKi 8.3–9.2) — a real preference, but not a clean separation. Intermedin, from the other gene, is less selective still: the 2004 paper that characterised it concluded it “represents a nonselective agonist for the RAMP coreceptors.”

Vasodilation is the famous action. Barrier integrity is the important one.

The 1993 discovery paper described a “potent and long lasting hypotensive effect,” and adrenomedullin is still routinely introduced as a vasodilator. The knockout mice tell a different story about what it is for.

• Adrenomedullin-null embryos do not survive. Caron and Smithies reported in 2001 that Adm−/− embryos “die at midgestation with extreme hydrops fetalis and cardiovascular abnormalities, including overdeveloped ventricular trabeculae and underdeveloped arterial walls.”
• Deleting RAMP2 reproduces it. RAMP2−/− embryos “died in utero at midgestation due to vascular fragility that led to severe edema and hemorrhage,” with endothelial cells detaching from the basement membrane and reduced expression of tight-junction, adherens-junction and basement-membrane molecules.
• Heterozygotes leak. In adult RAMP2+/− mice, reduced RAMP2 expression “led to vascular hyperpermeability and impaired neovascularization.”
• The lymphatics need it too. A companion study found that adrenomedullin-, calcrl– or RAMP2-null mice “died mid-gestation after development of interstitial lymphedema.”

Read together, these are not the phenotypes of a missing blood-pressure regulator. They are the phenotypes of a missing endothelial barrier. That reframing is what made adrenomedullin interesting in septic shock, where the defining problem is a vasculature that both dilates and leaks.

Adrenomedullin as a biomarker in sepsis and heart failure

Because adrenomedullin rises with the severity of vascular failure, it grades illness well. In a 2005 ICU cohort of 101 patients, median MR-proADM climbed stepwise from 0.4 nmol/L in healthy controls to 1.1 in SIRS, 1.8 in sepsis, 2.3 in severe sepsis and 4.5 in septic shock; non-survivors ran 8.5 versus 1.7 in survivors, with an area under the curve for survival of 0.81 — comparable to interleukin-6 and to the APACHE II and SAPS II scores.

The larger AdrenOSS-1 study measured bioactive adrenomedullin directly in 583 ICU patients with sepsis or septic shock. Median admission bio-ADM was 80.5 pg/mL and 28-day mortality was 22%. The more useful finding was about trajectory: among patients admitted above 70 pg/mL, those whose bio-ADM fell below that line by day 2 had 9.5% 28-day mortality, while those who stayed elevated had 38.1% (hazard ratio 4.9, 95% CI 2.5–9.8). That 70 pg/mL line then became the enrolment gate for the drug trial described below.

In acute dyspnoea, the 15-centre BACH study of 1,641 emergency-department patients found MR-proADM predicted 90-day survival in heart-failure patients with 73% accuracy versus 62% for BNP, and carried independent prognostic value in adjusted regression where BNP did not. Compare that with the natriuretic peptides, which remain the diagnostic test; adrenomedullin performs better as a prognostic one.

The drug that was designed not to block its target

The obvious way to drug adrenomedullin in septic shock would be to neutralise it, since it is elevated and it dilates vessels. The programme that actually reached patients does close to the opposite.

A 2013 mouse study compared anti-adrenomedullin antibodies matched for affinity but differing in epitope. The antibody directed at the peptide’s N-terminus substantially improved survival after cecal ligation and puncture (hazard ratio 0.077, 95% CI 0.019–0.315), and it “only partially inhibit[ed] ADM agonist activity in vitro.” The properly neutralising C-terminal binder did worse.

That N-terminal antibody became adrecizumab, later renamed enibarcimab. Its developers were explicit that “blocking of ADM does not accurately describe adrecizumab’s mechanism of action”: N-terminal binders “only marginally inhibit ADM activity, despite their high affinity and even when applied in vast molar excess over ADM,” and infusing the antibody produces a strong, dose-dependent increase in plasma adrenomedullin. The proposed explanation — and the authors label it a hypothesis, not a finding — is that antibody confined to the bloodstream drains small adrenomedullin molecules out of the interstitium into the circulation, raising barrier-stabilising signalling on endothelium while reducing dilatory signalling on the smooth muscle underneath.

What the human evidence shows

Elegant mechanism, unremarkable results. Every registered human study is charted below.

AdrenOSS-2 randomised 301 patients with septic shock and bio-ADM above 70 pg/mL to one of two adrecizumab doses or placebo. Its primary endpoints were safety and tolerability, and it met them: one infusion interruption, no haemodynamic disturbance, and grade 3 or higher treatment-emergent adverse events in 70.5% of the adrecizumab group versus 71.1% of placebo. On efficacy measures it was neutral. The difference in Sepsis Support Index was 0.72 (95% CI −1.93 to 0.49, p=0.24), and 28-day mortality was 23.9% versus 27.7% (HR 0.84, 95% CI 0.53–1.31, p=0.44). A change-in-SOFA analysis favoured treatment (difference 0.76, 95% CI 0.18–1.35, p=0.007). The authors’ own conclusion was that there were “no overt signals of harm… however, further randomized controlled trials are required to confirm efficacy and safety.”

ACCOST-HH tested a single 8 mg/kg dose in 150 patients with cardiogenic shock across four German university hospitals and was clearly negative: days free of cardiovascular organ support were 12.37 versus 14.05 (adjusted mean difference −1.69 days, p=0.37), with no mortality difference at 30 or 90 days. A phase 2 study in COVID-19 was terminated after 16 patients. A post-hoc AdrenOSS-2 analysis restricted to patients with low circulating DPP3 produced more favourable hazard ratios (0.61 and 0.49) but neither reached significance, and its authors flagged “all the caveats to be considered for post-hoc subgroup analyses.”

Native adrenomedullin has also been infused directly. A Japanese randomised phase 2a study gave 8-hour daily infusions for 14 days to 28 patients with steroid-resistant ulcerative colitis; the primary endpoint at 2 weeks showed no difference between the four groups. A reported 8-week remission difference of 100% versus 0% represents three patients and two patients respectively.

Searches of Drugs@FDA and DailyMed in late August 2026 return no approved application and no US prescribing label for adrenomedullin, adrecizumab or enibarcimab. A European orphan designation for a pegylated adrenomedullin in acute respiratory distress syndrome was withdrawn at the sponsor’s request in July 2024.

Handling and measurement notes

Adrenomedullin is not a compound anyone encounters as a vial in a fridge; it appears in the literature as an infused investigational agent or as an analyte. The general points below apply to peptides of this class rather than to adrenomedullin specifically.

• Amidation and the disulfide are load-bearing. A C-terminal amide and a small disulfide ring define activity across the whole calcitonin/CGRP family. Any handling that reduces disulfides or hydrolyses the amide destroys function without necessarily changing the molecular weight much — see how peptide sequences are written for why that ending is notated separately.
• Short circulating half-life is the norm. The reason MR-proADM exists as an assay target at all is that the active peptide is unstable in plasma. Related reading: half-life extension strategies.
• Concentration is a calculation, not a label. If you are converting between a stated mass and a working concentration for any peptide, the reconstitution calculator and the concentration converter do the arithmetic; solubility and diluent choice covers what dissolves in what.

Frequently asked questions

Is adrenomedullin the same as MR-proADM?

No. MR-proADM is residues 45–92 of the same precursor — a stable, biologically inert fragment used as a surrogate because it is easier to measure. It is produced in proportion to adrenomedullin, so it tracks it, but a result labelled MR-proADM is an inference about the hormone rather than a measurement of it.

Why is a drug for septic shock designed not to neutralise its target?

Because in animal models the non-neutralising, N-terminally directed antibody outperformed the neutralising one, and because adrenomedullin’s knockout phenotypes point to a protective endothelial-barrier role rather than a purely harmful vasodilatory one. Whether the redistribution hypothesis behind it is correct remains unsettled, and the clinical trials have not yet demonstrated benefit.

Is adrenomedullin related to CGRP?

Yes, in two senses. It is a member of the same calcitonin/CGRP peptide superfamily, and it signals through the same receptor protein, CLR — the difference being which RAMP is bound. Adrenomedullin also binds the CGRP receptor itself with only about an order of magnitude less potency than its preferred receptor.

Is any adrenomedullin drug approved?

Not in the United States. As of late August 2026, neither Drugs@FDA nor DailyMed lists an approved application or a prescribing label for adrenomedullin, adrecizumab or enibarcimab. The furthest any programme has progressed publicly is completed phase 2 work.

Informational only — not medical advice · 21+. Consult a qualified healthcare professional about any medical question.