Elunetirom is the first antidepressant candidate to act through brain thyroid-receptor signalling rather than monoamines — the reason it's worth watching. The catch: the only human efficacy signal comes from a single-arm, open-label study of ~30 patients with no placebo. Novel mechanism plus uncontrolled data is a hypothesis, not a near-term treatment.

Existing antidepressants and adjuncts work on monoamines or D2/5-HT targets. Elunetirom works somewhere new: it selectively activates thyroid hormone receptor β (TRβ) in the brain to drive energy metabolism and neuroplasticity.

• TRβ vs TRα — the precise version. Contrary to a common shorthand, TRα is actually the dominant thyroid receptor in the brain (~70–80% of brain TR), as well as in heart and bone; TRβ predominates in liver and kidney and is present in the brain only as the minority isoform. Elunetirom is engineered to hit TRβ specifically — the point is to spare TRα-mediated cardiac and skeletal effects, even though TRβ is not the brain's main receptor. The proposed antidepressant action is hypothesized to run through TRβ, not overall thyroid-receptor abundance.
• Prodrug. Elunetirom is converted to the active, brain-penetrant LL-340001.
• Downstream. Drives PGC-1α (mitochondrial energy), NRF2 (cytoprotection), and BDNF (plasticity).
• Preclinical readouts. ↑ neurogenesis, neurite outgrowth, synaptogenesis — labelled "psychoplastogenic" (company-derived term) and reportedly preserved under cellular stress.

Clinicians already know one version of thyroid-for-depression: T3 augmentation has decades of modest-evidence use, and supraphysiologic thyroid hormone has been tried in rapid-cycling bipolar illness. The limit was always that these used non-selective hormone, hitting cardiac and skeletal TRα — the dose-limiting problem (arrhythmia risk, bone loss).

Elunetirom's pitch: brain-targeted TRβ keeps the central effect and drops the peripheral cost. Whether that selectivity holds cleanly in humans is the open question, not a settled property.

Public data come mainly from a Phase 1 study in healthy volunteers (randomized, double-blind, placebo-controlled, single/multiple ascending dose): safe and well tolerated, no serious adverse events, dose-proportional PK, and CNS target engagement. Those results set Phase 2 dosing.

Three tiers — weakest where it matters most:

• Phase 1 (healthy volunteers). Randomized, double-blind, placebo-controlled. Solid — but a safety/PK study, not efficacy.
• Phase 2 AMPLIFY-BD (bipolar, NCT06869187). The efficacy headline (table below).
• Preclinical. The neuroplasticity work from Section 2.

The sponsor reports significance on the primary and all key secondary endpoints, a favourable safety profile, and a "rapid, robust, durable" effect. Full results: a medical conference in 2H 2026.

This section governs how much weight to give everything above. Per ClinicalTrials.gov, AMPLIFY-BD is single-arm, open-label, ~30 adults, all on background mood stabilizers and/or antipsychotics. No placebo, no randomization. That caps the numbers:

• No comparator = no drug-effect estimate. The 16.8-point drop bundles true drug effect, expectancy, natural recovery, regression to the mean, and background meds.
• Placebo response in bipolar depression is large — pooled placebo response rates run near 40% in controlled trials, with correspondingly big within-group score drops. Much of an uncontrolled 16.8-point change could be placebo and natural recovery, not drug.
• "Significant on all endpoints" is weak here — it means change-from-baseline differs from zero, not better-than-placebo.
• N ≈ 30, unblinded. Small and open-label — both inflate apparent effect.

None of this means it fails. It means we don't yet know, and the "rapid, robust, durable" language outruns the design that produced it.

On 26 May 2026 the FDA granted Fast Track for adjunctive bipolar I/II depression. It's routinely over-read.

It's procedural — a willingness to engage early, not a verdict on whether the drug works. Read it as "may move faster," not "works."

Tolerability looks favourable so far, but a 30-patient, 6-week open-label study can't answer the questions that matter for a chronic thyromimetic in bipolar patients. Track:

• Cardiac. The TRα-sparing thesis — watch for tachycardia, arrhythmia, or BP signals at scale; that's where imperfect selectivity shows first.
• Thyroid axis. TSH, free T4/T3, chronic-dosing effects.
• Bone. Thyroid excess drives bone loss; no long-term data exist.
• Liver — class precedent. TRβ selectivity is no guarantee of safety: eprotirome (KB2115), an earlier TRβ-selective agonist, was halted after liver-enzyme elevations in humans and cartilage damage in dogs. Resmetirom (Rezdiffra/MGL-3196), the first FDA-approved TRβ agonist (MASH, 2024), is liver-targeted and carries hepatic monitoring — a useful real-world reference point. Watch liver enzymes.
• Affective switch. Any antidepressant-acting adjunct raises hypomania/mania and cycling risk — uncharacterizable in 6 weeks at N≈30.
• Interactions. With mood stabilizers, antipsychotics, and thyroid-relevant drugs — undefined.

Keep every clause conditional. If an RCT confirms a real effect and tolerability holds, elunetirom slots in as an adjunct — added to mood stabilizers/antipsychotics, not monotherapy. The appeal is theoretical, based on mechanism rather than head-to-head data: a profile that might avoid the weight, metabolic, and movement liabilities driving discontinuation of current adjuncts, plus a distinct lever for partial responders. No comparative trial has shown any of this yet. Plausible niches: patients intolerant of antipsychotic adjuncts, partial responders, possibly anergic or cognitive presentations. All contingent on data that don't exist yet.

• Full AMPLIFY-BD results — conference, 2H 2026. Methods, secondaries, and safety detail the topline omits.
• AMPLIFY (MDD, NCT06633016) — topline Q3 2026; a key read-through to the larger population.
• The hurdle — a randomized, double-blind, placebo-controlled trial reproducing the signal. Until then, unproven.