KarX-EC

KarX-EC is Bristol-Myers Squibb's enteric-coated xanomeline capsule, co-administered with KarXT (the immediate-release M1/M4 muscarinic agonist combo approved as Cobenfy for schizophrenia in September 2024 [1]) to extend the same platform into Alzheimer's disease. Patients in the AD trials receive both: KarXT delivers immediate xanomeline plus peripheral cholinergic blockade via trospium, while KarX-EC adds a delayed release phase that bypasses the stomach. The combination is designed to smooth and extend systemic exposure while reducing the early-GI surge that drives nausea and vomiting. Four Phase 3 trials are recruiting: ADAGIO-1, -2, and -3 for agitation in AD [2][3][4], and MINDSET 2 for cognitive impairment [5]. The Cohen-Mansfield Agitation Inventory (CMAI-IPA) - a clinician-rated scale of agitation frequency and severity - is the primary efficacy endpoint in the ADAGIO keys. BMS paid roughly $14B for Karuna Therapeutics in early 2024 [6]; the AD program is the franchise extension that has to land for the deal math to clear its hurdle rate. ADAGIO topline readouts in 2027 are the gating events for the entire muscarinic platform thesis; MINDSET 2 reads out later, with study completion projected early 2029 [11].

Xanomeline isn't a new molecule. Eli Lilly studied it in AD in the 1990s and abandoned it after Phase 2 - oral dosing produced too much nausea, sweating, salivation, and vomiting to be tolerable [7]. Karuna's reformulation paired xanomeline with trospium, a muscarinic antagonist that doesn't readily cross the blood-brain barrier, so peripheral cholinergic side effects get blocked while central activity is preserved. That combination - KarXT - won FDA approval as Cobenfy in September 2024 for schizophrenia, the first novel mechanism in that indication in seven decades [1]. KarX-EC is the next formulation iteration: an enteric coat that delays xanomeline release until past the stomach, reducing upper-GI cholinergic effects further. In the AD trials, patients take KarXT and KarX-EC together - KarXT provides the immediate peak with trospium covering peripheral effects, and KarX-EC adds a delayed second wave of xanomeline absorbed in the small intestine. The intent is staggered, extended central exposure with a flatter peripheral side-effect curve. The exact AD-trial doses have not been publicly disclosed in detail and may be titrated differently than the schizophrenia Cobenfy regimen given the elderly population. This work is the foundation of BMS's bet that the muscarinic platform can stretch into elderly AD patients, who tolerate side effects worse than younger schizophrenia patients. Phase 3 program: ADAGIO-1 (NCT07011732) and ADAGIO-2 (NCT07011745) are mirror-image agitation keys, n=352 each, 14-week CMAI-IPA primary endpoint. ADAGIO-3 (NCT06937229) is the long-term safety extension at n=600. MINDSET 2 (NCT06976216) is the cognition study at n=586 with an ADAS-Cog11 endpoint and a 24-week treatment period. Combined enrollment across the four trials is ~1,890 patients. No public breakthrough or fast-track designation for the AD program. All four are listed as recruiting as of mid-2026.

Muscarinic receptors are like volume knobs on neurons - acetylcholine, the brain's attention and memory neurotransmitter, turns them up. There are five subtypes; M1 is dense in the cortex and hippocampus (memory and executive function), M4 is concentrated in the striatum (movement, reward, salience filtering). In Alzheimer's, the cholinergic neurons of the basal forebrain die off early and progressively. Donepezil, rivastigmine, and galantamine - the legacy AD drugs - work by blocking the enzyme that breaks down acetylcholine, propping up dwindling supply. Xanomeline takes the opposite angle: instead of preserving the neurotransmitter, it directly activates M1 and M4 receptors. For cognition, M1 activation in cortex is the rationale - boost signal in degraded memory circuits. For agitation, M4 activation in the striatum dampens hyperexcitable dopaminergic signaling, the same circuit logic that lets KarXT suppress psychotic symptoms in schizophrenia. The trospium question matters more in AD than it does in schizophrenia. Trospium is a quaternary amine designed not to cross the blood-brain barrier at standard doses, which is how Cobenfy gets away with blocking peripheral cholinergic side effects without compromising central xanomeline activity. But AD patients have compromised BBB integrity - chronic neuroinflammation, microvascular disease, and pericyte loss all increase paracellular leak. Even modest CNS trospium penetration could partially antagonize M1, the same receptor xanomeline is trying to activate. That's a mechanistic tension specific to this population, and BMS will need to demonstrate that CSF trospium remains negligible in elderly AD patients - or that any leak doesn't materially blunt cognitive endpoints. Regulators will look at this directly. How strong is the broader case? KarXT hitting the Positive and Negative Syndrome Scale (PANSS, the standard psychosis symptom rating) endpoint in EMERGENT-2 proved the M1/M4 axis can suppress psychotic symptoms in adults [8]. Bodick et al.'s 1997 xanomeline-monotherapy study in AD showed a cognitive and behavioral signal but couldn't sustain dose because of GI toxicity [7]. Extrapolation to AD agitation is mechanistically informed but not yet validated - agitation in AD is a heterogeneous neuropsychiatric syndrome, not the same neurochemistry as schizophrenia psychosis. The mechanism is plausible. It is not yet proven in this population.

ADAGIO-1 (NCT07011732) and ADAGIO-2 (NCT07011745) are mirror-image keys: randomized, placebo-controlled, n=352 each, 14-week primary endpoint on the Cohen-Mansfield Agitation Inventory-IPA version. Two identical Phase 3s is the standard FDA template for symptomatic CNS approvals - two independent positive trials is the historical bar, which is why Otsuka/Lundbeck needed BLAZE-1 and BLAZE-2 for Rexulti's 2023 AD agitation approval. BMS has not publicly disclosed any FDA agreement allowing one positive trial plus supportive data, so the operating assumption is that both ADAGIO-1 and ADAGIO-2 need to win. MINDSET 2 (NCT06976216) addresses the cognition question: n=586, 24-week treatment period, ADAS-Cog11 endpoint (the 11-item Alzheimer's Disease Assessment Scale-Cognitive subscale assessing memory, language, praxis, and orientation - the standard cognitive composite in AD trials). ADAGIO-3 (NCT06937229) is the long-term safety extension at n=600, tracking 12-month treatment-emergent adverse events. Design concerns: 14 weeks is short for an agitation endpoint, and placebo response in AD agitation trials typically runs 30-40% improvement at this duration. CMAI is clinician-rated, which introduces site-to-site variability. BMS is running parallel keys without a public Phase 2b specifically in AD agitation - the KarXT schizophrenia data is the bridging rationale. That's a confident bet but it leaves no early off-ramp. The placebo comparator is the right choice; Rexulti's adoption in AD agitation has been modest enough that there's no realistic active comparator. Enrollment is the rate-limiter - elderly AD recruitment timelines historically slip. MINDSET 2 study completion is projected around early 2029 per registry timelines [11], so cognition data lands well after the agitation readouts.

Our model gives this drug a 54% chance of eventually reaching approval. That number starts from the historical rate for Phase 3 drugs in this area - about 51% - then shifts based on ten details about the trial and its sponsor. The estimate gets a boost from a non-randomized design, more secondary endpoints than usual, and a light or open-label blinding approach; it is pulled back by weaker or limited results from earlier phases. The remaining factors fall close to average for this stage, so the final number stays near where the base rate started.

Efficacy risk, agitation: the CMAI-IPA endpoint is sensitive to placebo response, which runs 30-40% in AD agitation trials at 12-14 weeks. BMS needs a clinically meaningful delta, not just statistical significance - Rexulti's key effect size was roughly 3-4 CMAI points, and KarX-EC needs to match or beat that to justify the price premium. Efficacy risk, cognition: MINDSET 2's ADAS-Cog11 endpoint is the graveyard of AD trials. Lecanemab and donanemab succeeded by clearing amyloid; xanomeline's mechanism is symptomatic. A transient signal is the best plausible outcome and durability is unproven. Safety risk: KarXT in EMERGENT trials reported nausea (~19%), constipation (~21%), vomiting (~14%), and dyspepsia, with discontinuation rates around 7-9% [8]. The enteric coat is intended to reduce these but it's an unvalidated formulation strategy. In elderly AD patients, even modest GI burden translates to dropouts and caregiver fatigue. Trospium-in-AD is the subtler concern: BBB integrity declines with age and AD pathology, so the peripheral-restriction assumption that holds in schizophrenia patients is not automatically valid here. Any meaningful CNS trospium leak partially antagonizes the M1 activation the program is built on. Commercial risk: Cobenfy's launch has been notably slow. BMS reported $27M in Q1 2025 Cobenfy revenue [10], an annualized run-rate near $110M - well below the multi-hundred-million first-year consensus implied at acquisition. Payer prior-auth requirements and pricing around $22,500/year have constrained uptake. An AD agitation label expands the market but inherits the same payer dynamics. Rexulti is well-established in AD agitation since 2023 but its franchise revenue is dominated by MDD and schizophrenia; AD-agitation-specific revenue is materially smaller and not publicly broken out - the realistic indication ceiling is well below the franchise total. Execution risk: four parallel Phase 3s in elderly AD patients strain site networks; expect enrollment slippage and potentially asymmetric readouts across ADAGIO-1 and ADAGIO-2. IP/exclusivity risk: the underlying xanomeline composition-of-matter patents lapsed years ago. The commercial moat for KarX-EC rests on formulation patents (enteric coat, combination dosing) and regulatory exclusivity. Specific KarX-EC patent expiration dates are not publicly summarized in detail - this is a material gap for assessing the post-2030 commercial window.

Worth watching, signal-dependent. BMS paid $14B for Karuna in early 2024 [6] and the schizophrenia label alone clearly doesn't clear that hurdle - Cobenfy's $27M Q1 2025 print [10] makes that explicit. The Alzheimer's program is the franchise extension that makes the deal math work. If ADAGIO-1 reads out positive with a CMAI delta in the 3-4 point range and discontinuation rates below 10%, BMS has a multi-billion peak product and a validated muscarinic platform. If it misses or shows excessive GI dropouts, the KarX-EC enteric-coat thesis collapses and the Karuna deal looks worse retrospectively. The specific signal: paired readouts on ADAGIO-1 (NCT07011732) and ADAGIO-2 (NCT07011745) in 2027. One positive trial is hopeful; two positive trials is approval-ready. Absent any disclosed FDA agreement to accept one trial plus supportive data, the operating assumption is both have to win. Watch BMS earnings call commentary in late 2026 and 2027 for enrollment status and readout guidance. Secondary data to watch: real-world tolerability of Cobenfy in schizophrenia. If patients stay on it, the enteric coat is gravy; if discontinuation rates are high, KarX-EC matters more and the formulation strategy gets stress-tested in market data. Competitive context: Rexulti (Otsuka/Lundbeck) has held the AD agitation indication since 2023 but uptake has been modest. KarX-EC needs differentiated efficacy, cleaner safety, or both to displace it. The cognitive arm (MINDSET 2) is the swing factor - even a modest cognitive signal on ADAS-Cog11 would reshape the conversation and differentiate from a dopamine partial agonist. Cognition data does not land until ~2029 per registry timelines [11]. Check back: Q3 2026 earnings for enrollment update, ADAGIO topline expected 2027, MINDSET 2 readout ~2028-2029.