CORT108297

CORT108297 is a selective glucocorticoid receptor (GR) antagonist originally developed by Corcept Therapeutics and now being tested in two academic Phase 2 trials for stress-linked brain disorders: age-related memory decline at Johns Hopkins (NCT04601038, CORT-X, n=52) [1] and PTSD in military veterans at the San Francisco and Tuscaloosa VAs (NCT04452500, Seven Study, n=40) [2]. Neither is a Corcept-sponsored registration study. Johns Hopkins is the registered sponsor on CORT-X, the VA Office of Research and Development on Seven. Both are small, single-molecule proof-of-mechanism bets on whether dampening cortisol signaling in the brain can protect the hippocampus from stress-driven damage. Corcept has effectively parked this molecule behind its lead late-stage cortisol modulator relacorilant, and behind newer selective analogs like dazucorilant (ALS) and miricorilant (MASH, antipsychotic weight gain), so CORT108297's near-term commercial value is minimal [3]. The interesting question is scientific, not commercial. If the CORT-X pattern separation readout is positive, it says something real about GR antagonism as a brain-protective mechanism, which would matter for the whole class of CNS-penetrant GR antagonists, not just this compound.

Novel compound, never approved anywhere. Phase 2 in two indications, both academic-sponsored and small. CORT-X (NCT04601038) is sponsored by Johns Hopkins University, enrolls 52 older adults, and has an estimated primary completion date of November 2027 [1]. Seven Study (NCT04452500) is sponsored by the VA Office of Research and Development, enrolls 40 veterans across San Francisco and Tuscaloosa VA sites, and has an estimated primary completion date of December 2026 [2]. Total program enrollment across both trials is 92. No FDA breakthrough, fast track, or orphan designation is attached to either program. The CORT-X endpoint is a pattern separation memory task after two weeks of dosing, a mechanistic proof-of-concept measure rather than a clinical outcome [1]. The Seven Study primary endpoint is the CAPS-5 (Clinician-Administered PTSD Scale for DSM-5, a structured clinical interview scoring PTSD symptom severity that regulators accept as an efficacy endpoint), but at n=40 is powered only for large effects [2]. Both trials list "recruiting" status. Corcept is not listed as sponsor or collaborator on either. Corcept's own late-stage cortisol modulator, relacorilant, is Phase 3 in Cushing's syndrome and platinum-resistant ovarian cancer and consumes the pipeline attention and capital [3][4]. CORT108297 reads as a brain-penetrant back-catalog analog kept alive by outside investigators rather than an active commercial program.

Cortisol is the body's main stress hormone. It floods the bloodstream when you're threatened, sleep-deprived, or emotionally taxed, and it binds a protein called the glucocorticoid receptor (GR), which then travels to the cell nucleus and turns genes on or off [5]. Short bursts of this are useful. Chronic activation is corrosive, especially in the hippocampus, the brain region that files new memories, where sustained GR signaling shrinks dendrites, kills newborn neurons, and impairs pattern separation, the ability to tell similar memories apart. CORT108297 sits in the GR binding pocket and prevents cortisol from switching the receptor on. It has roughly 1000-fold higher selectivity for GR over the progesterone, estrogen, androgen, and mineralocorticoid receptors, unlike Corcept's approved GR blocker mifepristone (Korlym), which also hits the progesterone receptor and causes off-target endocrine effects [6][8]. CNS penetrance has been demonstrated in preclinical rodent models: CORT108297 crosses the blood-brain barrier and reduces hippocampal pathology after status epilepticus in mice, and shows behavioral effects consistent with central GR blockade [8]. No peer-reviewed human CSF or brain PET data are publicly available. The mechanistic case for CORT108297 in stress-driven cognitive decline is decent but not proven. Chronic cortisol elevation is a documented risk factor for cognitive decline in older adults, and CSF cortisol tracks with Alzheimer's disease progression. In rodents, GR antagonism reverses stress-induced memory deficits. But no GR antagonist has ever shown a cognitive benefit in a randomized human trial. Mifepristone's Cushing's approval validates GR as a druggable target for peripheral endocrine disease [6], not for brain outcomes.

Two active Phase 2 trials, both academic-run, both small. NCT04601038 (CORT-X): Johns Hopkins, 52 older adults at elevated Alzheimer's risk (26 with mild cognitive impairment due to AD, 26 cognitively normal at-risk), 2 weeks of CORT108297 versus placebo, primary endpoint is a pattern separation memory task, estimated primary completion November 2027 [1]. This is a fast, cheap, biomarker-adjacent proof-of-mechanism study, not a clinical efficacy trial. A positive signal would justify a larger, longer trial in prodromal AD (the pre-dementia stage where Alzheimer's pathology is accumulating but clinical symptoms are mild or absent) but would not itself support any regulatory path. NCT04452500 (Seven Study): VA Office of Research and Development, 40 veterans with PTSD across San Francisco and Tuscaloosa VA sites, CORT108297 versus placebo, primary endpoint is CAPS-5 score, estimated primary completion December 2026 [2]. Small, but the endpoint is regulator-recognized. The concerns are structural. Both trials are small and short. Neither has a commercial sponsor, so a positive academic readout has no automatic path to registration studies. Corcept's revealed preference (heavy investment in relacorilant, dazucorilant, and miricorilant, near-silence on CORT108297 in recent 10-Ks) suggests the company is not planning to pick up development on the back of these results [3][4]. Enrollment status is "recruiting" for both, and academic trial recruitment velocity in aging and PTSD populations is historically slow, particularly for cognitively-focused endpoints requiring extensive baseline neuropsych testing.

Our model estimates a 3% chance this drug is eventually approved. It starts from the historical base rate for Phase 2 drugs in this area (about 24%), then adjusts using ten facts about the trial and sponsor. What moves the number most: it is held back by heavier-than-usual blinding, the sponsor's thin or weak approval record, its few secondary endpoints, and weak or limited earlier-phase results. The other facts land near average for this stage, so they leave the estimate roughly where the base rate put it.

Efficacy risk is the biggest. Pattern separation is a legitimate hippocampal readout, but two weeks of dosing in healthy-at-risk older adults is an indirect measure of anything clinically meaningful. PTSD has been a graveyard for promising mechanisms (NMDA modulators, cannabinoids, ketamine follow-ons) that hit primary endpoints in small trials and failed in Phase 3. Safety risk is real. Mifepristone taught the class that GR blockade can cause adrenal insufficiency, hypokalemia (low potassium), and endometrial thickening [6]. CORT108297 is far more GR-selective than mifepristone and does not hit the progesterone receptor [8], which removes some issues, but chronic GR blockade still risks HPA-axis rebound (when you block cortisol's receptor peripherally, the brain senses a low cortisol signal and drives the adrenal glands to pump out even more cortisol, which can overwhelm the blockade and worsen the underlying problem) and metabolic effects. Phase 1 healthy-volunteer studies of CORT108297 reported adequate tolerability, but no detailed public safety database exists for chronic dosing [8]. In a healthy elderly population enrolled in a prevention-style trial, the safety bar is high. Execution risk: neither trial has a commercial sponsor. If Corcept stays disengaged, positive academic data may sit unmoved for years, and the compound could quietly die of orphan-asset neglect. Commercial risk: even a successful Phase 2 in AD prevention creates a difficult commercialization path. Prevention trials for cognitive decline are enormous (thousands of patients, multi-year follow-up, expensive imaging and cognitive batteries), and Corcept is a mid-cap company (market capitalization roughly $4 to $9 billion in 2026, revenue driven almost entirely by Korlym at $761 million in 2025) where near-term relacorilant approvals drive the investment case [3][4][9]. A back-burner brain-penetrant asset is not going to command that capital allocation absent a spectacular result.

Watch, do not weight. The interesting question is not whether CORT108297 succeeds commercially (it almost certainly will not on its own) but whether the CORT-X readout says anything real about GR antagonism as a brain-protective mechanism. If pattern separation improves on two weeks of dosing in older adults at AD risk, that is a mechanism signal worth tracking for the whole class, including any next-generation selective, brain-penetrant GR antagonists. The specific data point that turns this from noise into signal: a statistically clean positive on the CORT-X primary endpoint (NCT04601038) with a plausible effect size, ideally accompanied by a cortisol or HPA-axis biomarker showing target engagement. Datable catalyst checks: Seven Study estimated primary completion December 2026 (NCT04452500), CORT-X estimated primary completion November 2027 (NCT04601038). At n=52 and n=40 the data will either be clearly positive or clearly noise, no middle ground. In the meantime, the meaningful Corcept catalysts are relacorilant's Phase 3 readouts in Cushing's syndrome and platinum-resistant ovarian cancer, dazucorilant's planned Phase 3 in ALS (mid-2026), and miricorilant's Phase 2b readout in MASH (year-end 2026), all of which drive the near-term equity story and dwarf anything CORT108297 does [3][4]. For a Corcept thesis, this asset is a free option, not a reason to own the stock.