CORT108297

CORT108297 is a selective glucocorticoid receptor antagonist, meaning it blocks the receptor that cortisol (the body's stress hormone) binds to. Johns Hopkins is running a Phase 2 mechanistic trial (NCT04601038, n=52, randomized crossover, 120 mg/day for 2 weeks vs. placebo) testing whether short-course dosing improves hippocampal memory function in people aged 55+ either with mild cognitive impairment due to Alzheimer's disease or cognitively normal with family history, genetic risk, or subjective memory complaints [1]. The premise: chronic stress hormone exposure damages the hippocampus, the brain's memory hub, and an at-risk brain may be especially vulnerable. The compound originated at Corcept Therapeutics - the 'CORT' prefix is their naming convention - but Corcept's clinical focus is now on relacorilant, a related selective GR antagonist whose Phase 3 GRACE trial in Cushing's syndrome met its primary endpoint and which has a PDUFA date of July 11, 2026 in platinum-resistant ovarian cancer [2]. CORT108297 itself has become a tool compound for academic neuroscience rather than a registration-track drug. Treat this as signal-generating science. There is no commercial path to an Alzheimer's approval behind it.

Phase 2, recruiting, single-site academic trial at Johns Hopkins School of Medicine in Baltimore [1]. No FDA designations: no breakthrough therapy, no fast track, no orphan, no accelerated approval pathway. Johns Hopkins is the sponsor of record. That tells you the regulatory positioning - investigator-initiated science under NIH-style funding, with no industry-sponsored registration program behind it. A separate Phase 2a at the VA Office of Research and Development (NCT04452500, n=40 veterans, 180 mg/day for 7 days vs. placebo, primary completion Dec 31, 2026) tests the same compound in PTSD using the CAPS-5 scale (Clinician-Administered PTSD Scale, the standard clinician-rated severity instrument) [3]. Two small mechanism trials in adjacent stress-related indications. CORT108297 has been in academic hands for over a decade, appearing in published Corcept preclinical work in the early 2010s as a tool compound for separating glucocorticoid receptor effects from progesterone receptor effects [4]. Corcept itself has moved its commercial pipeline to newer assets, principally relacorilant for Cushing's syndrome (FDA issued a Complete Response Letter in late 2025 despite GRACE meeting its primary endpoint; the Phase 3 GRADIENT trial in adrenal Cushing's missed its primary endpoint) and platinum-resistant ovarian cancer (PDUFA July 11, 2026) [2]. CORT108297 has no announced corporate development program, no commercial readout timeline, and no expected NDA submission (New Drug Application - the filing made to the FDA to seek approval) in any indication. A clean positive Phase 2 mechanism signal here would, at best, justify a larger academic or industry-funded efficacy trial in a population sized for clinical endpoints.

Cortisol is the body's main stress hormone. It binds to the glucocorticoid receptor (GR), which then enters the cell nucleus and switches hundreds of genes on or off, affecting inflammation, metabolism, and brain function. In short bursts, this is useful - it's how your body responds to acute threats. Chronic exposure is the problem. Sustained high cortisol shrinks the hippocampus, the brain region responsible for forming new memories, and accelerates the kind of damage seen in Alzheimer's disease [5]. CORT108297 sits in the receptor's binding pocket and prevents cortisol from activating it, without hitting the progesterone receptor (which is what limits mifepristone, the first-generation GR antagonist sold as Korlym for Cushing's syndrome) [4]. The biological case is real. Cortisol-induced hippocampal damage is well-documented in animal models, observational human studies link cortisol levels to memory decline, and Cushing's patients (who have pathologically high cortisol) show reversible cognitive deficits when cortisol is brought down. Critical prior art: Corcept ran three Phase 3 trials of mifepristone in psychotic depression between 2004 and 2007, all of which failed to meet their primary endpoints [7]. That program is the most relevant precedent for the GR-antagonism-in-brain thesis, and it is why this target has remained academic rather than commercial. Selective compounds like CORT108297 were developed in part to remove off-target progesterone activity that complicated mifepristone's CNS trials. The leap is whether short-term selective GR blockade in genetically at-risk but cognitively intact people will produce measurable cognitive benefit on a two-week timescale. No selective GR antagonist has been approved for Alzheimer's or for any neuropsychiatric indication. The mechanism is plausible. The translation is unproven and has a documented history of clinical failure.

NCT04601038 is a single-site randomized double-blind crossover Phase 2 at Johns Hopkins, n=52 split into two arms of 26: (a) participants aged 55+ with mild cognitive impairment due to Alzheimer's disease, and (b) cognitively normal participants aged 55+ at elevated AD risk based on family history, genetics, and/or subjective memory complaints [1]. APOE4 carrier status is not listed as a hard inclusion criterion in the public protocol; risk enrichment is by family history and genetics broadly rather than strict APOE4 stratification. Dosing is 120 mg/day (2 tablets) for two weeks, with a matched placebo arm in crossover. The primary endpoint is pattern separation task performance, a hippocampal-dependent memory test where subjects distinguish similar but distinct stimuli, considered a sensitive early readout for Alzheimer's-related dysfunction. Primary completion date is listed as June 30, 2026, though academic timelines often slip. Two weeks of dosing is short: long enough to detect acute pharmacodynamic effects on memory, far too short to demonstrate disease modification. Sample size is small, and statistical power is limited to detecting moderately large effects on the pattern separation readout. The crossover design partially compensates for the small n by using each subject as their own control. This is the design of a mechanism-probing biomarker study, which is exactly what it should be at this stage. The companion PTSD trial (NCT04452500) at the VA uses CAPS-5 in n=40 veterans, 180 mg/day for 7 days vs. placebo, primary completion December 31, 2026 [3]. Neither trial is sized or designed to support an NDA. Recruitment is active at both sites per ClinicalTrials.gov. No interim analysis has been announced. A clean positive readout on pattern separation would be biologically interesting and would justify a larger study, but the path from there to an approved Alzheimer's drug is at minimum a decade and several hundred million dollars away.

Most drugs that enter Phase 2 never reach approval, and our model estimates this one has a 3% chance of eventually getting there. It starts from the historical approval rate for Phase 2 drugs in this area - about 24% - then adjusts based on ten specific facts about the trial and sponsor. The estimate is pulled down mainly by heavier-than-usual blinding, the sponsor's thin approval record, few secondary endpoints, and weak earlier-phase results. The remaining factors are close to average for this stage, so they don't move the number much either way.

Efficacy risk is largest. The mechanism is plausible but unvalidated in cognitive disorders - no selective GR antagonist has demonstrated clinical benefit in any CNS indication, and the most relevant prior program (mifepristone in psychotic depression) failed three Phase 3 trials [5,7]. The Hopkins trial enrolls cognitively intact at-risk people alongside MCI patients, which is the right early-prevention population but means a null result is hard to interpret: did the drug fail, or was the population too healthy to show effect? Two weeks of dosing may be too short to demonstrate hippocampal protection even if the mechanism works. The pattern separation endpoint is sensitive but its translation to clinically meaningful cognitive benefit is unproven. Safety risk is moderate. GR antagonism produces predictable on-target effects: disinhibition of the HPA axis (the hypothalamic-pituitary-adrenal feedback loop that regulates cortisol production - blocking the receptor removes negative feedback and drives cortisol higher), potential for adrenal insufficiency on withdrawal (when the adrenal glands cannot produce enough cortisol - clinically dangerous if dosing is stopped abruptly), and glucose dysregulation. Manageable in monitored trials but a real issue for any chronic dosing strategy in a healthy-ish population. Mifepristone's safety experience in Cushing's gives a useful benchmark, but Cushing's patients start from pathological cortisol levels; healthy at-risk subjects do not. Execution risk is meaningful at a single academic site with n=52 - slow enrollment, dropouts, and site-specific operational issues can all derail a small trial. Commercial risk is the cleanest answer: there is no commercial program. Corcept Therapeutics, the originator, has moved its development resources to relacorilant [2]. Even a clean positive Phase 2 readout here would need a sponsor - Corcept, NIH, or a new licensee - to fund a properly powered efficacy trial. None is currently announced.

Worth watching as science, not as a commercial bet. The signal to look for is a clean positive readout on the pattern separation task. That would be the first direct evidence that short-term selective GR blockade improves hippocampal function in at-risk humans, which is a genuinely interesting result worth following - particularly given the prior mifepristone failures in psychotic depression. The right place to check back is when the Johns Hopkins trial reports its primary outcome, currently listed as June 30, 2026, on ClinicalTrials.gov, though academic timelines tend to slip [1]. The companion PTSD trial at the VA has a primary completion of December 31, 2026 [3]. A coherent positive signal across both indications would be more meaningful than either alone. The commercial connection runs through Corcept Therapeutics (NASDAQ: CORT). Corcept reported Korlym revenue of $761.4 million in fiscal year 2025, up roughly 13% year-over-year, with 2026 guidance of $900 million to $1 billion [2]. Relacorilant's Phase 3 GRACE trial in Cushing's syndrome met its primary endpoint but received an FDA Complete Response Letter in 2025; the Phase 3 GRADIENT trial in adrenal Cushing's missed its primary endpoint; the platinum-resistant ovarian cancer NDA has a PDUFA date of July 11, 2026 [2]. Corcept does not discuss CORT108297 on earnings calls because it isn't a commercial asset. If the academic trials produce a clean positive signal and Corcept publicly resurrects CORT108297 or a backup compound for neuropsychiatric indications, that is the buy-side signal. Until then, this is biology to track, not a stock to trade. Check back in mid-to-late 2026 for the Hopkins primary completion and any Corcept pipeline update mentioning a neurology program.