A cortisol-blocking pill for Cushing's clears the FDA on one modest trial

Relacorilant, a pill that blunts the effect of the stress hormone cortisol, was approved by the US Food and Drug Administration in March 2026 for adults with endogenous Cushing's syndrome, a disease of chronic cortisol excess [s2]. The approval rests largely on one phase 3 trial, GRACE, in which patients who had already responded to the drug were more likely to keep their blood pressure under control if they stayed on it than if they were quietly switched to placebo — a real but narrow result from a trial that randomised only 62 people [s1].

Endogenous Cushing's syndrome is the state of prolonged exposure to too much cortisol, usually because of a hormone-secreting tumour of the pituitary or adrenal gland. The excess drives high blood pressure, high blood sugar, weight gain, muscle wasting, bone loss and mood disturbance. Surgery to remove the tumour is the first-line treatment, but it is not always possible or successful, and medicines are used when it is not. Relacorilant is a selective glucocorticoid receptor modulator: it competes with cortisol for the receptor cortisol acts through, so the hormone is still made at the same levels but its downstream effects are damped [s1].

Why a "selective" blocker matters

The company behind relacorilant, Corcept Therapeutics, already sells an older cortisol-receptor blocker, mifepristone, for the high blood sugar of Cushing's. Mifepristone is a blunt instrument: it also blocks the progesterone receptor, which is why the same molecule is used to end pregnancies and why it can cause irregular vaginal bleeding and endometrial thickening. The case for relacorilant is that it spares the progesterone receptor, and so should avoid those effects — a claim the trial's safety data are consistent with, as discussed below. The broader regulatory question hanging over government-owned cortisol dosing methods is covered separately in the dispute over an NIH mifepristone licence.

What the trial did

GRACE was a multicentre, double-blind, placebo-controlled, phase 3 study run at 77 centres across 11 countries and funded by Corcept [s1]. It used an unusual randomised-withdrawal design. First, all enrolled adults — aged 18 to 80, with hypercortisolism plus hypertension, high blood sugar, or both, and at least two signs or symptoms of the disease — took open-label relacorilant for 22 weeks, the dose escalated from 100 mg up to 400 mg once daily [s1]. Only patients who met predefined response criteria then entered the randomised phase, where they were assigned 1:1 to continue relacorilant or switch to placebo for 12 weeks, with neither patients nor investigators told which [s1]. The primary outcome was the proportion of patients who lost their blood-pressure response over those 12 weeks [s1].

The design is worth pausing on, because it flatters a drug. Everyone in the randomised comparison had already responded to relacorilant; the trial asked only whether stopping it caused a relapse, not whether starting it helps an unselected patient. It is a legitimate way to test a maintenance effect, but it is not the same as showing the drug works in the average person who walks into a clinic.

What it found

Of 404 patients screened, 152 were enrolled and started open-label relacorilant, and 95 completed that phase; 62 met the response criteria and were randomised — 30 to continue relacorilant (21 of whom had qualifying hypertension) and 32 to placebo (22 with qualifying hypertension) [s1]. Among those with baseline hypertension, significantly more of the patients switched to placebo lost blood pressure control than those kept on the drug: a proportion difference of 34%, with an odds ratio of 0.17 (95% confidence interval 0.04 to 0.77) and a P value of 0.022 [s1]. The wide confidence interval reflects how few patients that comparison rested on.

On safety, the events reported in the randomised phase were mostly musculoskeletal and mild. The most common in the relacorilant and placebo groups were back pain (5 [17%] versus 6 [19%]), acne (3 [10%] versus none), joint pain (3 [10%] versus 3 [9%]), bursitis (3 [10%] versus none), headache (3 [10%] versus 4 [13%]) and insomnia (none versus 4 [13%]) [s1]. Notably, the investigators recorded no cases of excessive cortisol-receptor blockade, adrenal insufficiency, vaginal bleeding with endometrial thickening, drug-induced low potassium, or QT-interval prolongation on the heart tracing [s1]. The absence of the endometrial and potassium signals is the concrete evidence behind the "selective" selling point.

How to read it

Relacorilant adds a second, more targeted cortisol-blocking option for a rare and serious disease where treatment choices are limited — and the trial genuinely shows that patients who respond keep a benefit while they take it [s1]. The honest caveats are the size and the design: the pivotal efficacy comparison involved a few dozen people who had already been selected as responders, and the primary endpoint was maintenance of a blood-pressure response rather than a hard clinical outcome such as fewer cardiovascular events [s1]. What the trial does not establish is how well the drug works when started in an unselected patient, or how it compares head-to-head with existing therapies, none of which GRACE included.

What to watch

The practical questions now are durability beyond 12 weeks, performance in the patients excluded by the response-enrichment design, and whether the cleaner receptor profile holds up in wider use [s1]. This article describes research and a regulatory decision and is not medical advice; treatment of Cushing's syndrome is a matter for specialist care.

Sources

• Efficacy and safety of relacorilant for the treatment of patients with Cushing's syndrome (GRACE) — The Lancet Diabetes & Endocrinology, 20 February 2026
• Drugs@FDA: Lifyorli (relacorilant), NDA 220641 — US Food and Drug Administration, approved 25 March 2026