BMS-986368

BMS-986368 (USAN: irafamdastat) is a dual FAAH/MAGL inhibitor that BMS is testing in Phase 2 for agitation in Alzheimer's patients (NCT06808984, n=120, placebo-controlled) [1]. The trial measures change on the Cohen-Mansfield Agitation Inventory, the standard scale clinicians use to count how often AD patients hit, scream, pace, or refuse care. Agitation affects roughly half of AD patients at some point in the disease course - an estimated 1-2M US patients with clinically significant symptoms. The only FDA-approved option today is Otsuka/Lundbeck's brexpiprazole (Rexulti), cleared in May 2023, which posts modest efficacy and inherits the antipsychotic-class black box warning for elevated mortality in elderly dementia patients [4]. Otsuka/Lundbeck have publicly forecast ~$1B al annual Rexulti revenue from the AD agitation indication, framing the commercial opportunity [9]. The competitive picture has shifted: Axsome's AXS-05 (dextromethorphan-bupropion) sNDA was accepted with Priority Review and a PDUFA date of April 30, 2026, on the back of three positive Phase 3 trials (ADVANCE-1, ACCORD-1, ACCORD-2) [10]. AXS-05 will almost certainly be approved before BMS-986368 finishes Phase 2, meaning BMS-986368 is targeting a two-drug approved market, not a one-drug market. BMS-986368 takes a mechanistically different swing: instead of tweaking dopamine and serotonin (brexpiprazole) or NMDA/sigma-1 (AXS-05), it boosts the brain's endocannabinoid system - the same signaling system that cannabis interacts with. BMS is running a parallel Phase 2 in multiple sclerosis spasticity (NCT06782490, n=200) [2]. The class carries scar tissue. The 2016 BIA 10-2474 Phase 1 incident in France killed one healthy volunteer and severely brain-damaged four others [5]. Forensic analysis pinned that on off-target activity rather than FAAH inhibition itself, but every FAAH program since has run on a regulatory tightrope. Watch the Phase 2 safety package as closely as the efficacy delta.

Novel small molecule, never approved anywhere. Two active Phase 2 trials are recruiting: AD agitation (NCT06808984, n=120) [1] and MS spasticity (NCT06782490, n=200) [2], both measuring symptomatic endpoints over weeks. Three Phase 1 studies have completed - healthy volunteers (NCT06411730, n=56, completed November 2024) [3], drug-drug interaction with itraconazole plus food/pH effects (NCT06170723, n=32), and radiolabeled ADME (NCT06227975, n=8). BMS has not posted Phase 1 results to ClinicalTrials.gov and has made no conference presentation or peer-reviewed publication of Phase 1 safety, PK, or selected Phase 2 dose. That absence is itself a data point worth flagging: for a FAAH-class compound entering a behaviorally vulnerable elderly population, the lack of any public Phase 1 safety disclosure means external observers cannot independently assess whether BMS-986368 cleared the class-specific neurological AE bar that BIA 10-2474 established. The Phase 1 elderly cohort (within NCT06411730) is particularly relevant given the Phase 2 AD population. No FDA breakthrough therapy, fast track, orphan drug, or accelerated approval designation has been announced. The compound came through Celgene's pipeline; sponsor records on ClinicalTrials.gov still read Celgene, which BMS acquired in 2019 for $74B [7]. Public Phase 2 readout dates have not been disclosed. The AD agitation trial started recruiting in 2025 with a 12-week behavioral endpoint, so a realistic readout window is late 2026 to mid-2027 (modeled here as 2027-Q1). The MS spasticity readout likely trails 6 to 12 months behind given the larger enrollment target. Neither trial is paired with a biomarker companion diagnostic; enrollment criteria are clinical (CMAI severity for AD, modified Ashworth scale for MS). BMS has not called the asset out on recent quarterly disclosures (Q1 2025 10-Q filed April 24) [6], which is consistent with a non-core CNS program progressing quietly through Phase 2 while the company's external narrative stays focused on oncology and cardiometabolic.

FAAH (fatty acid amide hydrolase) and MAGL (monoacylglycerol lipase) are the two enzymes that break down the body's own cannabinoids. FAAH degrades anandamide. MAGL degrades 2-arachidonoylglycerol. Block both and you raise endocannabinoid signaling without putting any THC into the system. The premise for AD agitation: stronger endocannabinoid tone calms hyperactive amygdala circuits and dampens behavioral disinhibition. Cannabinoid receptors sit densely in brain regions that go haywire in AD behavioral symptoms. Mechanism validation is mixed. Single-target FAAH inhibitors (Pfizer's PF-04457845) and MAGL inhibitors have run through pain, PTSD, anxiety, and Tourette syndrome - generally clean safety, no efficacy wins. PF-04457845 failed Phase 2 in PTSD on the primary endpoint [8]. No FAAH or MAGL inhibitor has been approved for any indication, anywhere. Open Targets evidence flags FAAH inhibition primarily for migraine and pain rather than behavioral or cognitive endpoints, which is a gap. The class also carries the BIA 10-2474 scar: a Bial FAAH inhibitor killed one healthy volunteer and severely brain-damaged four others in a French Phase 1 in 2016 [5]. Forensic work pinned it on off-target activity (the molecule hit multiple enzymes beyond FAAH), not the mechanism itself. Regulators have moved more cautiously on every FAAH program since. The dual FAAH/MAGL bet is that hitting both endocannabinoid arms together does what neither alone could. No clinical proof of that exists yet. Competitive mechanism context: AXS-05 (the imminent regulatory competitor) targets NMDA receptors and dopamine/norepinephrine reuptake with sigma-1 agonism - a mechanistically distinct path. Its positive Phase 3 results validate the AD agitation endpoint as druggable but say nothing about endocannabinoid efficacy in this population.

NCT06808984, Phase 2, recruiting, n=120, sponsor still listed as Celgene (BMS subsidiary) [1]. Placebo-controlled. Primary endpoint is change from baseline in CMAI total score - a 29-item scale clinicians use to count and rate agitation episodes including hitting, screaming, pacing, sexual disinhibition, and refusing care. This is the same endpoint Otsuka used for brexpiprazole's AD agitation approval, so the regulatory path is mapped and FDA expectations are calibrated [4]. The 120-patient size fits Phase 2 dose-finding and signal detection. It is not a registrational trial; BMS will need a much larger Phase 3 if the Phase 2 reads positive. No biomarker selection: patients enroll on clinical agitation severity, which is appropriate for a behavioral symptom but means the trial cannot identify a molecular responder subgroup if the overall effect is mixed. Enrollment criteria filter on mild-to-moderate AD with clinically significant agitation requiring caregiver confirmation; specific CMAI threshold and MMSE/CDR staging cutoffs are not detailed in the public ClinicalTrials.gov record. The companion MS spasticity trial (NCT06782490, n=200) uses the modified Ashworth scale, a clinically validated muscle tone metric used in approved MS spasticity drugs (baclofen, tizanidine) [2]. Running both Phase 2 trials in parallel is a sensible strategy for an endocannabinoid asset: it tests the mechanism in two distinct CNS settings (behavioral and motor) and gives BMS two shots on goal. Both trials began recruiting in early 2025. No interim analysis has been disclosed. The trial design itself is conventional and adequately powered for the chosen endpoints. The open questions are about the mechanism and the missing Phase 1 readout, not the protocol.

Our model puts this drug's chance of eventual approval at 11%. That starts from a 24% historical baseline for Phase 2 drugs in this area, then adjusts based on ten facts about the trial and sponsor. The estimate gets a boost from an above-average number of secondary endpoints and a strong sponsor track record, but is pulled down by weak earlier-phase results and heavier-than-usual blinding. The remaining factors land near average and leave the number close to where the base rate set it.

Safety is the headline risk. The 2016 BIA 10-2474 incident traced to off-target activity, not FAAH inhibition itself, but FDA and EMA both run FAAH programs more cautiously now [5]. BMS-986368 needs a clean safety package across the three completed Phase 1 studies and the active Phase 2. The fact that BMS has not yet publicly disclosed Phase 1 safety data (no ClinicalTrials.gov results posting, no conference abstract, no press release) is a soft negative - investors and clinicians have no external read on whether the elderly cohort tolerated the compound. Any neurological adverse event in Phase 2 - even transient - will get magnified by reviewers and the press. Efficacy risk is real because the bar is rising. Brexpiprazole won FDA approval in May 2023 with a roughly 30% reduction on CMAI vs placebo [4]. AXS-05 hit its primary endpoint in three Phase 3 trials and is on Priority Review with a PDUFA date of April 30, 2026 [10]. By the time BMS-986368 reads out (modeled 2027-Q1), it will most likely be entering a two-drug approved market, not a one-drug market. BMS-986368 needs to match brexpiprazole on efficacy with a cleaner safety profile (atypical antipsychotics carry the elderly-dementia mortality black box) or beat both incumbents on efficacy. The dose-response curve for endocannabinoid modulation in elderly brains is unforgiving: too much tone causes sedation, cognitive blunting, and falls, which are catastrophic in this population. Commercial risk: the AD agitation market is real but contested. Otsuka/Lundbeck have forecast ~$1B additional Rexulti revenue from the indication [9]; analyst peak US sales estimates for a differentiated entrant typically sit in the ~$1-2B range. Payers will compare cost to generic risperidone (used off-label, pennies per pill) and to Rexulti (~$1,400 per month US list). With AXS-05 likely the second approved entrant and BMS-986368 chasing the third slot, a novel mechanism without dramatic separation on safety or efficacy hits step-therapy walls. Execution risk is low for BMS - they can finish the trials. The program is non-core to their strategic priorities (oncology, hematology, cardiometabolic). A pipeline reshuffle, an M&A move, or a missed quarterly number could push agitation in AD down the priority list [6].

Worth tracking. Not a signal yet, and the competitive backdrop just got harder. The mechanism is biologically novel for AD agitation, the indication has real unmet need (estimated 1-2M US patients with clinically significant symptoms, brexpiprazole the only currently-approved option carrying the antipsychotic-class mortality warning) [4], and BMS has the resources to push a winner through Phase 3. But AXS-05's Priority Review and April 30, 2026 PDUFA date [10] mean BMS-986368 is now chasing a two-drug approved market by the time it has Phase 2 data. The class graveyard is real, BMS has not flagged this as a strategic priority on recent earnings calls [6], and BMS has not made any Phase 1 safety data public. The single data point that turns this into a signal: a Phase 2 CMAI readout showing 30%+ improvement vs placebo with no neurological AEs in elderly patients, plus a clean delta vs the AXS-05 safety/tolerability profile that will exist by then. Anything weaker and the program likely gets shelved, partnered, or sold. BMS's company revenue runs ~$48B annually [6]. A CNS Phase 2 asset is rounding error on the P&L. Track this for class validation, not for BMS earnings impact. Analyst peak US sales estimates for a differentiated AD agitation entrant fall in the ~$1-2B range; a clean Phase 2 here would reopen interest in endocannabinoid programs across pain, anxiety, PTSD, and movement disorders, where smaller biotechs have parked similar molecules. A Phase 2 failure would likely close the class for another funding cycle. Check back: AXS-05 PDUFA action on/around April 30, 2026; BMS Q3 and Q4 2025 earnings calls for any pipeline disclosure or Phase 1 data drop; then the AD agitation Phase 2 full readout (most likely 2027-Q1 to 2027-Q3). The MS spasticity readout will probably trail 6 to 12 months. If BMS adds a third Phase 2 indication or an FDA designation, that's the early tell that internal data looks better than the public picture suggests.