A working map of how second-generation (atypical) antipsychotics are sequenced in practice, how their receptor profiles predict their side-effect signatures, and where the newest mechanisms — muscarinic agonism chief among them — are reshaping the algorithm for the first time in seventy years.
Published in The Lancet Psychiatry in March 2025, INTEGRATE is the first consensus algorithmic guideline assembled by authors from every UN region, built from an umbrella literature review, expert workshops, and lived-experience focus groups. Its central move: stop treating "first-generation vs. second-generation" as the operative category, and route treatment by receptor profile, side-effect burden, and response timeline instead. Click a step to expand it.
Select the first agent by side-effect profile and shared decision-making, not by assumed efficacy — head-to-head efficacy differences between most SGAs are modest, with clozapine the clear outlier. Establish a metabolic baseline before the first dose, not after weight gain appears.
Assuming confirmed adherence, the first antipsychotic should run at a therapeutic dose for at least four weeks before response is judged. Reaching for a switch or an add-on earlier than this window mistakes a slow responder for a non-responder.
Responding → continue, move to maintenance-phase monitoring and consider earlier LAI conversion if adherence is a concern.
Persistent positive symptoms → switch to a pharmacodynamically distinct agent (e.g., a D2 full antagonist → a partial agonist, or vice versa) rather than another drug from the same functional family. Shared decision-making around side-effect trade-offs happens again at this branch.
Same rules apply: confirm adherence, hold at therapeutic dose, reassess at four weeks. Side-effect mitigation (metabolic, EPS, prolactin) continues in parallel — it is not deferred to "after" the efficacy question is settled.
Non-response to two adequate trials of different antipsychotics, at therapeutic dose, with confirmed adherence, defines TRS. INTEGRATE's key departure from older algorithms: move to clozapine promptly at this point rather than cycling through additional D2-blocking agents first. Delaying clozapine is one of the most consistently documented sources of prolonged, avoidable disability in TRS.
Positive, negative, and cognitive symptoms are treated as separate domains needing separate strategies, rather than assuming one dose increase addresses all three. Negative and cognitive symptoms respond far less reliably to dose escalation than positive symptoms do, and over-titrating against them mainly adds side-effect burden.
Psychiatric prescribers are expected to own physical-health monitoring, not defer it to primary care by default — nonadherence is driven overwhelmingly by side-effect burden, especially metabolic and sexual side effects patients may not spontaneously report. Substance-use comorbidity is treated as an expected complicating factor to screen for, not an exception.
Twelve SGAs in routine use, filterable by mechanism. Bars show relative receptor impact (qualitative — teaching-level, not literal Ki values) across the five axes that predict most of what you'll see clinically: D2 (efficacy + EPS + prolactin), 5-HT2A (the "atypical" antagonism that softens EPS), H1 (sedation + weight), M1 (anticholinergic burden), and α1 (orthostatic risk).
Select up to four agents to overlay their receptor-impact profiles on one radar. This is the fastest way to eyeball why two drugs that both "work" produce completely different side-effect conversations with a patient — compare olanzapine against aripiprazole, or clozapine against cariprazine, to see the shape of the trade-off.
For seventy years, "antipsychotic" meant D2 blockade or partial agonism. That changed in September 2024. The agents below are current to mid-2026.
Cobenfy (xanomeline–trospium, formerly KarXT) — FDA approved September 2024, the first antipsychotic with no D2 blockade at all. Xanomeline is a dual M1/M4-preferring muscarinic agonist; trospium is a peripherally restricted muscarinic antagonist added specifically to blunt the cholinergic GI and autonomic side effects that made xanomeline unusable as monotherapy on its own. Because it never touches D2, it sidesteps EPS, prolactin elevation, and — per EMERGENT trial data and the 52-week pooled safety results presented at APA 2026 — much of the classic metabolic burden of dopamine-blocking agents.
Real-world data presented at APA's May 2026 annual meeting associated it with reduced hospitalizations and lower overall antipsychotic burden; separate switch-study data (SIRS 2026, Florence) showed symptom stability in patients transitioned directly from oral atypicals. Practically: think of it first for patients whose EPS or metabolic tolerance has been the limiting factor on conventional SGAs, not as a universal first-line replacement — the evidence base and clinician familiarity are both still young relative to the D2-blocking class.
Watch: cholinergic GI effects (nausea, dyspepsia, constipation, salivation) even with trospium on board — these are the main tolerability-limiting side effects reported. More than seven companies now have distinct M1/M4-agonist compounds in development, so this becomes a real subclass, not a one-off.
INTEGRATE pushes LAI conversion earlier — as a proactive adherence strategy, not a last resort after repeated relapse. The pipeline here has been unusually active through 2025–2026, especially around closing the gap for olanzapine, which has long lagged risperidone/paliperidone/aripiprazole in injectable options.
| Molecule | Brand (LAI) | Interval | Status | Notes |
|---|---|---|---|---|
| Risperidone | Risperdal Consta | Every 2 weeks | established | Original IM microsphere LAI; requires oral overlap during the first ~3 weeks. |
| Risperidone | Perseris | Monthly | established | Subcutaneous, no oral overlap needed. |
| Risperidone | Uzedy | Monthly or every 2 months | established | Subcutaneous. Label expanded to bipolar I maintenance — for that indication use the monthly schedule only; the 2-month interval is not approved for bipolar maintenance. |
| Risperidone (generic) | Amneal risperidone ER | Monthly | new · 2025 | FDA-approved Sept 2025 with 180-day Competitive Generic Therapy exclusivity; launch expected late 2025. |
| Paliperidone | Invega Sustenna | Monthly | established | Requires a same-day + day-8 loading protocol at initiation. |
| Paliperidone | Invega Trinza | Every 3 months | established | Requires ≥4 months stable on Sustenna first. |
| Paliperidone | Invega Hafyera | Every 6 months | established | The longest interval of any antipsychotic LAI; requires stability on Sustenna or Trinza first. Invega's LAI family is the deepest of any molecule in the class. |
| Aripiprazole | Abilify Maintena | Monthly | established | Requires 14 days of oral overlap at initiation. |
| Aripiprazole | Abilify Asimtufii | Every 2 months | established | First 2-month LAI antipsychotic indicated for both schizophrenia and bipolar I maintenance monotherapy in the US. |
| Olanzapine | Zyprexa Relprevv | Every 2–4 weeks | established, limited use | Carries a boxed warning for post-injection delirium/sedation syndrome (PDSS) — mandates observation for 3 hours post-injection at a certified facility, which has sharply limited real-world uptake. |
| Olanzapine | TEV-'749 (investigational) | Monthly (planned) | pipeline · NDA filed 2025 | Subcutaneous, Medincell SteadyTeq copolymer delivery. NDA submitted Dec 2025, accepted by FDA Feb 2026. Phase 3 SOLARIS data through week 56 showed no PDSS signal — if approved, this would be the first practical, low-observation-burden olanzapine LAI. |
The standard ADA/APA-derived monitoring cadence, applied to any SGA — heavier for higher-metabolic-risk agents (olanzapine, clozapine, quetiapine), lighter but not skipped for lower-risk agents (aripiprazole, lurasidone, ziprasidone, cariprazine, brexpiprazole).
● baseline draw ● scheduled check annual reassessment of the full panel continues indefinitely thereafter.
A rough starting-titration shape for a subset of agents — illustrative only, always defer to current prescribing information and the patient in front of you.
Severe neutropenia — mandatory absolute neutrophil count (ANC) monitoring: weekly for 6 months, biweekly for the next 6 months, then monthly indefinitely while on therapy. As of late 2025, US pharmacies no longer require REMS enrollment to dispense clozapine, and prescribers no longer need to register patients in a federal database — but the ANC monitoring obligation itself is unchanged and remains mandatory.
Myocarditis (highest risk in the first 4–8 weeks), seizure risk (dose-dependent, especially above ~600 mg/day), and severe constipation / ileus — the anticholinergic burden from clozapine's high M1 affinity makes GI motility a frequently underweighted monitoring item relative to the more heavily emphasized hematologic risk.
INTEGRATE's central practical recommendation on this front is timing: initiate clozapine promptly once TRS criteria are met — two adequate trials, confirmed adherence, inadequate response — rather than after additional D2-blocking trials. Delayed clozapine initiation is one of the most consistently reproduced findings in the TRS literature as a driver of avoidable long-term disability, and it remains the only antipsychotic with demonstrated superiority specifically in treatment-resistant illness.