An Oral Relaxin Mimic Changed Heart-Failure Hemodynamics. The Hard Endpoint Is Still Missing.

A pill that copies one part of a pregnancy hormone's biology has produced an intriguing signal in chronic heart failure. In the phase 2b LUMINARA trial, AZD5462 lowered vascular resistance across three tested doses in one cohort and moved a measure of cardiac remodeling in the right direction in another.

That sounds like a clean win. It was not.

The strongest remodeling result missed the usual threshold for statistical significance by a sliver, and neither primary endpoint tells us whether patients lived longer, avoided hospital, or felt better. LUMINARA is important because it takes an unusual peptide-hormone pathway into a sizable randomized human trial. It is also a good example of why mechanism, surrogate endpoints, and patient benefit must be kept in separate boxes.

The news: relaxin biology, rebuilt as a daily pill

AZD5462 is a once-daily oral agonist of relaxin family peptide receptor 1, usually shortened to RXFP1. The natural ligand, relaxin-2, is an insulin-like peptide hormone best known for helping the cardiovascular and connective-tissue systems adapt during pregnancy. RXFP1 activation can reduce systemic vascular resistance and cardiac afterload, support organ perfusion, and influence tissue remodeling.

There is one important naming detail. AZD5462 is not itself a peptide. It is a small molecule engineered to activate a receptor normally controlled by a peptide hormone. That distinction matters for anyone following peptide therapeutics: this is not oral delivery of relaxin. It is an attempt to reproduce selected relaxin signaling without asking a fragile peptide to survive the gut or requiring repeated injections.

The primary LUMINARA results appeared online in Circulation on August 30, 2026. Investigators enrolled adults already receiving strong guideline-directed heart-failure therapy, so AZD5462 was tested as an addition rather than a replacement.

What LUMINARA actually tested

LUMINARA was an international, double-blind, placebo-controlled, dose-ranging trial conducted at 57 sites in 10 countries. Participants received placebo or 20, 80, or 360 mg of AZD5462 once daily for 24 weeks. Randomization was 1:1:1:1.

The study was really two related experiments.

Cohort A included 235 people with a left ventricular ejection fraction of 35% or lower. Its primary endpoint was change in end-systolic volume index, a body-size-adjusted measure of how much blood remains in the left ventricle after contraction. A lower value can indicate favorable reverse remodeling.

Cohort B included 140 people with an ejection fraction from 41% to 55%. Its primary endpoint was change in systemic vascular resistance index, a measure of the resistance the heart pumps against.

The groups were not especially diverse by sex. Mean age was 65 in cohort A and 70 in cohort B; men accounted for 88% and 69% of the cohorts, respectively. That imbalance does not erase the findings, but it limits confidence about how well the results generalize.

The result worth watching, and the decimal point that changes it

In cohort A, the highlighted 20 mg dose produced a placebo-adjusted change in end-systolic volume index of -5.4 mL/m² at week 25. The 95% confidence interval ran from -10.9 to 0.1 mL/m², with P=0.054.

The direction is encouraging. The confidence interval, however, crosses zero, and the P value sits just above the conventional 0.05 cutoff. Calling this a proven remodeling benefit would overstate the evidence. Calling it meaningless because it missed by 0.004 would also be too glib. The honest reading is that LUMINARA found a borderline signal that needs confirmation.

The dose pattern deserves scrutiny too. The abstract spotlights 20 mg for the remodeling result, despite testing 80 and 360 mg as well. Without complete dose-by-dose data, it is impossible to know whether the lower dose was genuinely optimal, whether higher exposure flattened the benefit, or whether random variation influenced which arm looked best. More dose does not always produce more benefit when a receptor has several downstream signaling routes.

Cohort B delivered the cleaner statistical result. All three AZD5462 doses significantly reduced systemic vascular resistance index compared with placebo at week 25, with P values below 0.05. The abstract does not provide the exact reduction for each dose, so it would be wrong to invent a magnitude or declare one dose superior.

This finding supports the proposed mechanism: RXFP1 activation appears capable of reducing the resistance facing the heart in people with chronic heart failure. But a hemodynamic change is not the same as a clinical outcome. Blood vessels can behave differently on a monitor while hospitalization risk, exercise tolerance, symptoms, kidney function, and survival remain unchanged.

Why the mechanism is more interesting than another vasodilator story

Heart-failure treatment already includes drugs that affect vascular tone, fluid balance, kidney signaling, heart rate, and remodeling. AZD5462 is interesting because relaxin biology may connect several of those systems at once.

Earlier translational work reported that long-term AZD5462 treatment in a non-human-primate model of heart failure reduced end-systolic volume by roughly 40% and increased ejection fraction by 20% after 84 weeks, without significant changes in blood pressure or heart rate. Phase 1 human work was described as generally well tolerated and showed pharmacologic responses consistent with RXFP1 activation.

Animal numbers that dramatic can create unrealistic expectations. Experimental heart failure is not human heart failure, and 84 weeks in a controlled primate model is very different from decades of vascular disease, diabetes, kidney dysfunction, arrhythmia, and medication use in real patients. LUMINARA matters precisely because it tests whether that biology leaves a detectable trace in humans receiving modern care. It did, but the trace was mixed.

Structural work adds another layer. A 2026 bioRxiv preprint used cryo-electron microscopy to compare RXFP1 bound to native relaxin-2 with RXFP1 bound to AZD5462. The researchers reported that the hormone engages the receptor's outer domain, while AZD5462 binds within the transmembrane region and stabilizes a distinct active state. In plain English, both turn on the same receptor, but they do not press the same molecular button.

That could explain why a small-molecule mimic does not reproduce every effect of the natural hormone. It could also make dosing less intuitive: a receptor can route signals through different cellular pathways depending on how it is activated. The structural paper is a preprint and should be treated as mechanistic context, not settled clinical evidence.

Safety: reassuring wording, incomplete detail

The Circulation abstract says AZD5462 was well tolerated compared with placebo. That is useful, especially for a therapy intended to sit on top of several existing heart-failure drugs. It is not a complete safety profile.

The abstract does not provide treatment-emergent adverse-event counts, discontinuations by dose, clinically important hypotension, kidney events, electrolyte changes, or rhythm findings. Those details matter when the drug's purpose is to change vascular resistance. A favorable top-line description should not be converted into “safe,” especially after 24 weeks and 375 randomized participants.

The study also has an industry context readers should see clearly. AstraZeneca developed AZD5462, and company employees appear throughout the author list. The lead author is affiliated with the Baim Institute, Massachusetts General Hospital, and Harvard Medical School, while the collaboration spans academic and clinical sites. Sponsorship does not invalidate a randomized trial. It does make independent replication, transparent dose-level data, and outcome-focused follow-up more important.

What would turn this signal into a heart-failure treatment?

The next trial needs to ask questions patients can feel. Does AZD5462 reduce worsening-heart-failure events or cardiovascular death? Does it improve exercise capacity or a validated health-status score? Are the hemodynamic effects durable? Which ejection-fraction group benefits, and which dose produces a consistent response without hypotension or kidney trouble?

A larger study should also enroll more women. Relaxin biology has obvious sex-related context, while LUMINARA's reduced-ejection-fraction cohort was overwhelmingly male. That mismatch is not a side note.

Finally, the development program must show that the borderline remodeling signal was not a lucky arm in a multi-dose experiment. A prespecified dose, replicated endpoint, coherent dose-response story, and patient-level outcomes would make the case far stronger.

The Oria take

AZD5462 is one of the more thoughtful examples of the peptide-adjacent drug space. Instead of trying to force a peptide hormone into a convenient formulation, researchers built an oral small molecule around its receptor. LUMINARA shows that the concept is pharmacologically active in humans with chronic heart failure.

What it does not show is that AZD5462 improves how long or how well those patients live. One primary endpoint produced statistically significant hemodynamic changes; the highlighted remodeling result narrowly missed significance. Both are reasons to continue the work, not reasons to call the problem solved.

For now, the right headline is neither “failed” nor “breakthrough.” A novel receptor strategy crossed from elegant biology into a randomized human signal. The next trial has to cross the harder distance from signal to benefit.

Sources

Januzzi JL Jr et al. Oral Relaxin Receptor Agonist AZD5462 in Participants With Chronic Heart Failure: Primary Results From the LUMINARA Trial. Circulation. 2026. PMID 42668430. https://pubmed.ncbi.nlm.nih.gov/42668430/

ClinicalTrials.gov. LUMINARA, NCT06299826. https://clinicaltrials.gov/study/NCT06299826

Quintana-Hayashi MP et al. Phase 2b trial of an oral relaxin family peptide receptor 1 agonist in patients with chronic heart failure: rationale and design. ESC Heart Failure. 2026. PMID 41711202. https://pubmed.ncbi.nlm.nih.gov/41711202/

Ryberg E et al. The first oral relaxin receptor RXFP1 agonist for heart failure treatment: Translational studies from non-human primates to humans. Molecular Therapy Advances. 2026. PMID 42137267. https://pubmed.ncbi.nlm.nih.gov/42137267/

Osei-Owusu J et al. Divergent activation of the RXFP1 relaxin receptor by protein and small molecule agonists. bioRxiv preprint. 2026. PMID 42327029. https://pubmed.ncbi.nlm.nih.gov/42327029/

**Evidence grade: B-.** LUMINARA was a randomized, double-blind, placebo-controlled phase 2b trial with 375 participants and prespecified physiologic endpoints. Confidence is limited by short follow-up, surrogate rather than clinical outcomes, a borderline remodeling result, incomplete abstract-level safety and dose data, male-heavy enrollment, and the need for independent confirmation.

**Medical disclaimer:** This article is for education only and does not provide medical advice, diagnosis, or treatment. AZD5462 is investigational and is not an approved heart-failure therapy. Do not change prescribed heart-failure medication or seek investigational compounds outside a registered clinical trial. Oria exists to make emerging peptide and peptide-adjacent research easier to understand, not to replace care from a qualified clinician.