Every antipsychotic approved in the seventy years since chlorpromazine's accidental discovery has shared one mechanism: blocking dopamine D2 receptors. The atypical antipsychotics — clozapine, risperidone, quetiapine, olanzapine — added 5-HT2A antagonism and offered a somewhat different side effect profile, but the dopamine blockade remained the presumed active mechanism. KarXT's September 2024 FDA approval broke that streak. It works through muscarinic acetylcholine receptors and doesn't touch dopamine at all. That's not a minor pharmacological variant — it's the first genuinely new mechanism in the history of antipsychotic drug development, and it opens a research agenda that the field hasn't had before.
This article is for informational purposes only and does not constitute medical advice. Clinical trial eligibility and availability vary. Always consult a qualified healthcare professional before making any medical decisions or considering participation in a clinical trial.
Summary
Seventy years of antipsychotic pharmacology produced drugs that reduce positive symptoms — hallucinations and delusions — through dopamine D2 blockade, while doing relatively little for the negative symptoms and cognitive impairment that account for most of schizophrenia's functional disability. KarXT (xanomeline-trospium, marketed as Cobenfy by Bristol Myers Squibb), approved by the FDA in September 2024, targets muscarinic M1/M4 receptors instead of dopamine — the first mechanistically distinct antipsychotic ever approved. In 2026, muscarinic follow-on compounds, NMDA receptor-enhancing agents, and long-acting injectable programs are driving the most active research pipeline in this space in decades.
ClinicalMetric Analysis
- Cobenfy's negative symptom improvement is pharmacologically significant — prior antipsychotics improved negative symptoms primarily as artifacts of positive symptom reduction, not via direct mechanism, and Cobenfy's M1-mediated prefrontal effect is different in kind. D2-blocking antipsychotics reduce hallucinations and delusions, which reduces the psychotic burden that indirectly drives negative symptoms. Cobenfy's negative symptom improvement at effect sizes not previously seen in Phase 3 suggests direct prefrontal M1 mechanism action. Confirming whether this translates to sustained functional improvement — employment, social engagement, independent living — requires longer-term data than EMERGENT-3's 24-week extension provides. The functional outcome trials are the clinically decisive evidence still pending.
- The tardive dyskinesia hypothesis — that non-D2 mechanisms reduce TD risk over time — requires multi-year follow-up to confirm, and prescribers switching patients to Cobenfy for this reason should document the rationale and track outcomes. The mechanism prediction is sound: TD results from dopamine receptor supersensitivity after chronic D2 blockade. Without D2 blockade, the supersensitivity pathway isn't activated. But this remains a mechanistic prediction, not a clinical observation. EMERGENT-4 long-term safety data and post-market pharmacovigilance will establish actual TD rates. Prescribers should not represent reduced TD risk as established without the long-term data — but they should document TD risk concerns in patients they switch, because this creates the observational record that will validate or refute the hypothesis.
- Long-acting injectable formulation development for Cobenfy or follow-on muscarinic agents is the most clinically impactful next step — adherence is the largest single modifiable driver of schizophrenia hospitalization. LAI dopaminergic antipsychotics dramatically reduce relapse rates and hospitalization compared to oral formulations, primarily through adherence improvement. If muscarinic agents can be formulated as monthly or quarterly injections, the adherence benefit combines with the mechanistic advance — and the combined effect on functional outcome would be substantially larger than either improvement alone. BMS has disclosed LAI development interest; the timeline, formulation approach, and Phase 1 start date are the key near-term data points to monitor.
KarXT: How a Failed 1990s Drug Became a Breakthrough
Xanomeline was originally developed at Eli Lilly in the early 1990s as a potential Alzheimer's treatment — it was a muscarinic M1/M4 receptor agonist with promising cognitive effects in preclinical work. Clinical trials found it reduced positive and negative schizophrenia symptoms, which was not the intended target but was striking. The problem: severe peripheral side effects. Nausea, vomiting, hypersalivation, and excessive sweating from muscarinic receptor activation in the gut and other peripheral tissues made the drug intolerable and the program was abandoned.
Karuna Therapeutics' critical insight was to combine xanomeline with trospium chloride — a muscarinic antagonist that doesn't cross the blood-brain barrier. Trospium blocks the peripheral muscarinic receptors responsible for the GI side effects while xanomeline's central M1/M4 agonism remains active. The EMERGENT-1 and EMERGENT-2 Phase 3 trials demonstrated that this combination (now Cobenfy) significantly reduced total PANSS scores versus placebo: by 9.6 points in EMERGENT-2 and 11.6 points in a confirmatory analysis — improvements that exceeded the standard threshold for clinical meaningfulness (typically defined as 6–10 PANSS points). Critically, both positive symptom subscale (PANSS positive) and negative symptom subscale (PANSS negative) improved, at effect sizes not previously seen in a Phase 3 trial for negative symptoms. No weight gain. No significant EPS. No metabolic changes. BMS acquired Karuna for $14 billion in 2024.
Why M1/M4 Agonism Works in Schizophrenia: The Mechanism That Makes This Interesting
This is worth understanding because it explains why follow-on muscarinic programs are a credible research avenue rather than opportunistic copycat development.
The muscarinic M1 receptor is heavily expressed in the prefrontal cortex and hippocampus — the same circuits that govern working memory, executive function, and attentional control. These regions are hypoactive in schizophrenia and are the anatomical substrate of the cognitive impairment that most determines functional outcome. Post-mortem studies of schizophrenia brains consistently find reduced muscarinic receptor expression, implying a cholinergic deficit as a core feature of the disorder rather than an epiphenomenon. M1 agonism in the prefrontal cortex enhances cognitive function directly by improving the signal-to-noise ratio in working memory circuits.
The M4 receptor, expressed in the striatum, modulates dopamine release indirectly — M4 agonism reduces dopamine neurotransmission via a mechanism that doesn't involve D2 blockade. This likely explains Cobenfy's antipsychotic effect on positive symptoms: it's not that it doesn't affect dopamine; it affects dopamine differently, upstream and through a modulatory rather than blockade mechanism. Whether this distinction translates to better long-term outcomes — less tardive dyskinesia, better adherence — is exactly what post-marketing and long-term safety studies are tracking.
Muscarinic Follow-On Compounds and the Emraclidine Program
Eli Lilly — which originally developed xanomeline and watched Karuna prove the concept — launched emraclidine, a highly selective M4 receptor agonist with a different binding mode than xanomeline. The selectivity matters: M4-only activity might provide antipsychotic effects with a cleaner central profile than mixed M1/M4 agonism. Phase 2 data for emraclidine (LY3839379) in schizophrenia showed statistically significant PANSS total score reductions compared to placebo at 6 weeks — not as large as Cobenfy's Phase 3 signal, but adequate to advance to Phase 3. Lilly is funding a full Phase 3 program.
AbbVie and Johnson & Johnson both have muscarinic agonist programs in early clinical development. The hypothesis driving all of them is the same: if the cholinergic deficit is a core feature of schizophrenia rather than a consequence, targeting it should produce effects that dopamine blockade cannot — particularly on negative symptoms and cognition. We don't yet know if that hypothesis fully holds, but the Phase 2 data from multiple compounds now support it well enough to justify Phase 3 investment.
The Glutamate Hypothesis and NMDA-Targeting Drugs
The NMDA receptor hypofunction hypothesis of schizophrenia — the idea that reduced NMDA-mediated glutamate signaling in interneurons drives the cognitive and negative symptom syndrome — has been generating drug programs for decades with mixed success. The challenge has been that direct NMDA agonists risk excitotoxicity, and most approaches to enhancing NMDA function have shown modest or inconsistent clinical effects.
Iclepertin (BI 425809, Boehringer Ingelheim)
A glycine transporter 1 (GlyT1) inhibitor that increases extracellular glycine — a co-agonist required for NMDA receptor activation. Phase 2 results (NCT02847559) showed a 1.1-point improvement on the MATRICS Consensus Cognitive Battery (MCCB) composite score versus 0.1 for placebo at 12 weeks. The effect is modest by most standards, but the MCCB is a demanding composite measure, and any movement in the cognitive impairment associated with schizophrenia (CIAS) is considered clinically relevant. The Phase 3 CONNICS trial enrolled over 900 patients; results are expected in 2025–2026 and will be a key data readout for the NMDA hypothesis.
Luvadaxistat (TAK-831)
A D-amino acid oxidase (DAAO) inhibitor that raises D-serine levels — another NMDA co-agonist — by blocking D-serine catabolism. Phase 2 data in schizophrenia showed cognitive improvements on the MCCB and a reduction in negative symptom severity on the BNSS scale, advancing this compound to Phase 3 planning. Different targets for essentially the same hypothesis; whether one proves more effective than the other will depend on which node in the pathway is more tractable clinically.
Negative Symptoms: The Hardest Problem
Negative symptoms — blunted affect, alogia, avolition, asociality, anhedonia — drive most of schizophrenia's functional disability, yet no drug is specifically approved for them. All current antipsychotics were approved for positive symptoms; any improvement in negative symptoms is a secondary benefit, typically modest, and often confounded by improvement in secondary negative symptoms (those caused by positive symptoms, depression, or medication side effects).
Roluperidone (MIN-101, Minerva Neurosciences) is a sigma-2 receptor antagonist and 5-HT2A antagonist with no dopamine D2 activity — developed specifically to target primary negative symptoms. The Phase 3 CYPRESS trial failed its primary endpoint in the overall population but showed statistically significant benefit in a pre-specified subgroup of patients with predominantly negative symptoms (PANSS positive score ≤18, negative factor score ≥20). An enriched Phase 3 recruiting this specific population is underway. The target enrichment strategy is methodologically sound and increases the likelihood of a clean positive result — the question is whether FDA will accept an enriched population approval.
Long-Acting Injectables: Underutilized and Underinvestigated
The FIRST trial (NCT02919189) randomized first-episode schizophrenia patients to aripiprazole lauroxil long-acting injectable (LAI) versus oral antipsychotic as initial therapy. Patients randomized to LAI had significantly better outcomes at 24 months on every measure that mattered: fewer hospitalizations, better functional outcomes, higher treatment continuity. This is the most important clinical trial result in LAI research in years, and it hasn't penetrated clinical practice adequately. Most patients still start with oral medication despite evidence that LAI as first-line therapy produces better outcomes in the early critical period.
In 2026, the LAI pipeline has extended to six-month formulations (paliperidone palmitate 12-month, Invega Hafyera) and once-yearly products in development at Teva. Research is also examining whether LAI antipsychotics in ultra-high-risk individuals — those meeting clinical high-risk criteria for psychosis conversion — can delay or prevent transition to full schizophrenia. These trials are small but scientifically important.
Key Takeaways
- KarXT (Cobenfy) reduced PANSS total scores by 9.6–11.6 points versus placebo in Phase 3, improving both positive and negative subscales — without D2 blockade, weight gain, or metabolic effects. The mechanism is real and is being replicated in follow-on programs.
- Emraclidine (M4-selective agonist) showed Phase 2 efficacy with a different receptor binding profile; Lilly's Phase 3 program will clarify whether M4 selectivity offers clinical advantages over mixed M1/M4 agonism.
- Iclepertin's Phase 3 CONNICS results in 2025–2026 will be a pivotal test of whether NMDA receptor enhancement via GlyT1 inhibition provides clinically meaningful cognitive improvement at scale.
- The FIRST trial established that LAI antipsychotics as first-line therapy produce better outcomes than oral medication in first-episode schizophrenia — this finding is not yet adequately reflected in clinical practice.
- Roluperidone's enriched Phase 3 in patients with predominantly negative symptoms (PANSS positive ≤18) is the best-designed attempt to date to specifically treat the most refractory aspect of schizophrenia.