What Is Relaxin? The Hormone That Keeps Failing Its Trials

Peptide science

Relaxin has one of the most frustrating track records in modern cardiovascular medicine. The hormone was discovered in 1926, and on paper it does almost everything a heart-failure drug should do: it widens blood vessels, raises kidney blood flow, and pushes back against scar tissue. A century later, no relaxin-based medicine has been approved anywhere. This guide walks through what relaxin is, what it genuinely does, and why three separate large trials have now come up short.

What is relaxin?

Relaxin belongs to the insulin superfamily, and it is built like insulin rather than like a simple peptide strand. Humans carry three non-allelic relaxin genes. RLN2 encodes the hormone that actually circulates in blood, produced mainly by the corpus luteum of the ovary; RLN1 is expressed in tissues such as prostate but not placenta; and RLN3 is expressed essentially only in the brain. Most other mammals have a single gene equivalent to human RLN2.

Mature relaxin-2 is a two-chain protein of roughly 6 kDa: a 24-residue A-chain and a 29-residue B-chain, held together by two interchain disulfide bonds plus one disulfide inside the A-chain. It is translated as a much longer precursor with a 108-residue connecting C-chain that is cut out during processing. The structure was solved by X-ray crystallography at 1.5 Å resolution in 1991 (PDB entry 6RLX).

The pregnancy story, and the ligament claim people get wrong

Frederick Hisaw named the hormone in 1926 after injecting serum from pregnant animals into virgin guinea pigs and watching the pubic ligament relax within six to eight hours. That single experiment is the origin of the popular idea that relaxin loosens ligaments throughout pregnancy.

Two things complicate it. First, the species matter: pigs, guinea pigs and rodents show a relaxin surge at the end of gestation, which is where the cervix-softening story comes from. Humans and other primates show no such near-term surge, and the hormone’s role in human labour is described in the literature as uncertain. In women, relaxin appears in the late luteal phase, rises sharply after conception, peaks at the end of the first trimester, and then settles to intermediate levels.

Second, the human joint data do not support the attribution. Schauberger and colleagues reported in 1996 that peripheral joint laxity increases in pregnancy but does not correlate with serum relaxin levels. Marnach and colleagues measured wrist laxity in 46 women each trimester in 2003: about 54% showed at least a 10% increase from first to third trimester, but the changes did not track with maternal relaxin, estradiol or progesterone. A 2012 systematic review of relaxin and pregnancy-related pelvic girdle pain screened 731 references, included six studies, and rated the evidence for an association as low.

Four receptors, two very different architectures

The relaxin family peptide receptors, RXFP1 through RXFP4, split into two groups. RXFP1 and RXFP2 are leucine-rich-repeat receptors (formerly LGR7 and LGR8) with a large horseshoe ectodomain plus a small LDLa module that is found nowhere else in this receptor class and is essential for signalling; splice variants carrying only the LDLa module behave as antagonists. RXFP3 and RXFP4 are ordinary short-N-terminus peptide GPCRs.

Coupling differs accordingly. RXFP1 and RXFP2 signal through Gs, GαoB and Gαi3, raising cAMP, while RXFP3 and RXFP4 couple to Gi/o and lower it. RXFP1 additionally drives MAP kinase, PI3K and nitric-oxide signalling. A 2023 cryo-EM structure of active human RXFP1 with Gs concluded that the receptor works through an unusual mechanism of autoinhibition.

What relaxin actually does

Cardiovascular and renal adaptation. In pregnancy, systemic vascular resistance falls steeply in the first trimester, cardiac output rises by roughly 40%, and mean arterial pressure drops only 5 to 10 mmHg. Renal plasma flow increases 40 to 65% and glomerular filtration rate increases 50 to 85%. The causal evidence in animals is strong: relaxin-neutralising antibodies or ovariectomy abolish gestational renal vasodilation in conscious pregnant rats. In humans, women who conceive with donor eggs have no corpus luteum and no circulating relaxin, and their first-trimester GFR rise is markedly subdued.

Anti-fibrotic signalling. Relaxin acts through RXFP1 to phosphorylated ERK1/2, then neuronal nitric oxide synthase, nitric oxide and cGMP. It upregulates matrix metalloproteinases and inhibits the TGF-β1 / pSmad2 axis. In TGF-β1-stimulated human dermal fibroblasts, MMP-1 rose about 50% and MMP-2 and MMP-9 about 80%.

Relaxin-3 in the brain. Relaxin-3 is made mainly by GABAergic neurons of the nucleus incertus, which project to hypothalamus, hippocampus, amygdala and prefrontal cortex. It has been implicated in feeding, stress responses, arousal and cognition.

INSL3 and RXFP2. The related peptide INSL3 drives outgrowth of the gubernaculum during the transabdominal phase of testicular descent. Mice lacking either Insl3 or Rxfp2 show bilateral cryptorchidism as essentially their only phenotype, and mutations in both genes have been reported in patients with cryptorchidism.

Serelaxin: the trial that looked like a win

Serelaxin is recombinant human relaxin-2. A phase 2b dose-finding study in acute heart failure (Pre-RELAX-AHF, 234 patients) selected 30 µg/kg/day for phase 3. RELAX-AHF, published in The Lancet in 2013, randomised 1,161 patients to a 48-hour infusion or placebo within 16 hours of presentation.

The result was genuinely mixed. The first co-primary endpoint, dyspnoea relief measured as visual-analogue-scale area under the curve to day 5, was met (448 mm × h, p = 0.007). The second co-primary, Likert-scale dyspnoea improvement over the first 24 hours, was not: 26% on placebo versus 27% on serelaxin, p = 0.70. Readmission outcomes showed no effect. What generated the excitement was a prespecified additional endpoint: deaths at 180 days, 65 on placebo versus 42 on serelaxin, hazard ratio 0.63, p = 0.019. That was a mortality signal from a drug given for two days.

Regulators were not convinced by one study. An FDA advisory panel voted unanimously against approval in March 2014 and a Complete Response Letter followed in May. The EMA’s CHMP issued a negative opinion in January 2014, confirmed the refusal on re-examination in May, and formally refused marketing authorisation for Reasanz in August 2014, citing the missed 24-hour endpoint, the unclear clinical relevance of the day-5 benefit, imputation of values for patients who had died or needed rescue treatment, and reliance on a single main study.

RELAX-AHF-2 settled it. Published in the New England Journal of Medicine in 2019 with 6,545 patients, the same drug at the same dose failed both primary endpoints: cardiovascular death at 180 days was 8.7% versus 8.9% (p = 0.77), and worsening heart failure at day 5 was 6.9% versus 7.7% (p = 0.19). An accompanying editorial argued that the earlier vasodilator mortality reductions had been spurious.

Why the biology has not translated

Pharmacokinetics are part of the story. Published half-life estimates for serelaxin differ by source, ranging from roughly two hours to under five hours, and serum levels approach steady state within about four hours of starting an infusion and fall rapidly on stopping. A 48-hour exposure was being asked to move a 180-day endpoint.

Other indications fared no better. A 231-patient phase 3 in diffuse cutaneous systemic sclerosis found no difference in modified Rodnan skin score at any timepoint, forced vital capacity declined in the relaxin groups, and stopping treatment at week 24 produced significant falls in creatinine clearance with serious renal adverse events in seven relaxin-treated patients versus none on placebo. A randomised trial of recombinant relaxin for cervical ripening did not advance ripening. In preeclampsia, only a phase 1 safety study was ever run; there is no positive human efficacy trial.

Where relaxin research stands in 2026

The most important recent readout is volenrelaxin (LY3540378), a long-acting relaxin engineered with an albumin-binding domain and given as a weekly subcutaneous injection. Its phase 2 trial in worsening HFpEF was stopped early by the sponsor after 332 participants because of evidence of worsening congestion. The 25 mg dose did improve left-atrial reservoir strain (+3.9%, p = 0.006) but 50 mg and 100 mg did not, NT-proBNP rose 24.5% across pooled doses, and heart-failure hospitalisation trended the wrong way (HR 2.64, p = 0.070).

AstraZeneca removed AZD3427, an Fc-fusion relaxin mimic, from its pipeline in February 2026 after underwhelming phase 2 efficacy in 260 people with heart failure. What remains is AZD5462, described as the first small molecule targeting relaxin biology to enter clinical trials, with phase 2b data in chronic heart failure expected in the second half of 2026. In the laboratory, ML290 remains an important tool compound: the first RXFP1-selective small-molecule agonist, biased in its signalling and, notably, active at human, monkey and pig RXFP1 but not the mouse receptor.

Handling and measurement notes

• It is not a linear peptide. Relaxin-2 is a two-chain, disulfide-linked heterodimer. It cannot be synthesised, refolded, quantified or stored the way a single-chain synthetic peptide can.
• Production is a refolding problem. Early material was made by synthesising both chains in reduced form and oxidatively refolding them. Modern recombinant routes express a single chain with a short connecting linker in E. coli inclusion bodies, refold it, then remove the linker enzymatically.
• Oxidation liabilities. Protein-engineering work has flagged two methionine residues in the B-chain as oxidation-prone, which is relevant to any discussion of storage and degradation.
• Concentration arithmetic still applies. If you are converting between mass, volume and concentration for any peptide preparation, the reconstitution calculator and concentration converter handle the maths.

Frequently asked questions

Does relaxin cause loose joints in pregnancy?

Joint laxity does increase during pregnancy, but multiple human studies have failed to find a correlation between the degree of laxity and circulating relaxin levels, and a systematic review rated the evidence linking relaxin to pelvic girdle pain as low. The popular version of this claim outruns the data.

Is relaxin approved as a medicine anywhere?

No. Serelaxin, the recombinant form, was refused marketing authorisation by the EMA in 2014 and received a Complete Response Letter from the FDA the same year. No relaxin-based product has been approved since.

What is the difference between relaxin-2 and relaxin-3?

Relaxin-2 is the circulating hormone made by the corpus luteum and acts mainly at RXFP1. Relaxin-3 is expressed almost exclusively in the brain, principally in the nucleus incertus, and acts mainly at RXFP3. They are different genes with different jobs despite the shared family name.

Why did serelaxin have to be given as a 48-hour infusion?

Because its half-life is measured in hours, not days. Levels approach steady state within about four hours and fall rapidly once the infusion stops, so continuous intravenous delivery was the only way to maintain exposure. That mismatch between a two-day drug and six-month endpoints is one of the standard explanations offered for the negative outcome trial.

Informational only — not medical advice · 21+. VialHelp does not sell peptides and does not recommend any product, brand or protocol. Trial results described here are reported for context and are not recommendations. Consult a qualified healthcare professional for personal medical questions.