Published on 2 October 2026
Augmenting an antidepressant with an antipsychotic after four weeks of inadequate response: what does the perospirone trial show?
The essentials
In this double-blind randomised trial run in ten Chinese hospitals, 210 patients with a moderate to severe major depressive episode, who had improved too little after at least four weeks of an SSRI or SNRI, received either perospirone or placebo on top of their antidepressant for eight weeks. The publication sets two primary outcomes at eight weeks. Remission is more frequent with perospirone, 55.2% versus 38.9% (adjusted odds ratio 1.944; 95% CI 1.060 to 3.564; p = 0.032). The difference in response is not significant, 67.8% versus 60% (p = 0.28). No correction was applied for these two outcomes, the difference in remission is no longer significant in either sensitivity analysis, and the authors themselves describe their result as hypothesis-generating. The clearest separation appears at four weeks. Perospirone may not be authorised where you practise (in France, for example, no medicinal product containing it is listed in the public medicines database): the value of the study lies in the question of when to augment an antidepressant, and in what it teaches about reading a trial.
Context
When a patient responds poorly to an adequately conducted antidepressant treatment, the question is not only what to add but also when. Waiting eight weeks before changing anything prolongs an episode that is already long. According to the authors, several guidelines, including CANMAT, recommend reassessing the antidepressant after four weeks and accept adding a second-generation antipsychotic in case of inadequate response (Lam et al., 2024).
Perospirone is a second-generation antipsychotic, first approved in Japan for schizophrenia. According to the authors, its use as an augmenting agent in depression had so far rested on one published case and one open-label study.
The study at a glance
| Item | |
|---|---|
| Population | |
| Patients with a moderate to severe major depressive episode (MADRS of at least 20) and a reduction of less than 50% after at least four weeks of an SSRI or SNRI: 210 randomised, 177 in the full analysis set | |
| Intervention | |
| Perospirone 4 to 48 mg per day (less than 4 mg in a few patients), most often between 4 and 16 mg, added to the SSRI or SNRI, for eight weeks | |
| Comparator | |
| Placebo added to the SSRI or SNRI | |
| Outcomes | |
| Two primary outcomes at week 8: response (MADRS reduction of at least 50%) and remission (MADRS of 10 or less). Secondary outcomes: response and remission at week 4; changes in MADRS, QIDS-SR16, HAMA and Q-LES-Q-SF at weeks 4 and 8. In version 2.0 of the protocol (16 December 2021), response and remission at week 4 were also primary outcomes; an undated adjustment reclassified them as secondary. Other secondary scales planned in the protocol (GAD-7, PSQI, SHAPS, CGI, BSI-CV, SDS, cognitive tests) are not reported | |
| Design | |
| Phase IV, multicentre trial (ten hospitals), 1:1 randomised with stratification by centre, double-blind and placebo-controlled; main analysis by logistic regression adjusted for baseline MADRS, with missing data replaced by last observation carried forward; mixed model for repeated measures for score changes; registered with the Chinese Clinical Trial Registry (ChiCTR2200063354) | |
Quality check
| Item | Judgement |
|---|---|
| Randomisation and blinding | Double-blind |
| Finding1:1 randomisation against placebo. The integrity of blinding was not tested. | |
| Dropouts | About 20% |
| FindingTwice the 10% anticipated in the published sample size calculation; the authors consider that power may have been reduced as a result. The analysis population (177 patients) remains close to the 176 required by that calculation. | |
| Multiplicity | Not corrected |
| FindingTwo primary outcomes, with no adjustment. A p value of 0.032 on one of the two is not enough to draw a conclusion. | |
| Sensitivity analyses | Remission not confirmed |
| FindingMultiple imputation and per-protocol analysis, in addition to the main model. The difference in remission at eight weeks is no longer significant in either (p = 0.12 in both cases); the four-week results remain significant. | |
| Independence | Worth reading |
| FindingNo conflict of interest declared. The trial is sponsored by the Livzon company, which manufactures perospirone and supplied the trial drug and placebo. Reboscience, a private company, is among the funders and carried out the statistical analysis. The authors state that the funders and sponsor had no role in the design and conduct of the study, the collection, management and interpretation of the data, or the preparation of the manuscript; analysis does not appear on that list. | |
Results
| Item | Value |
|---|---|
| Remission, week 8 | 55.2% vs 38.9%; adjusted OR 1.944 (95% CI 1.060 to 3.564); p = 0.032 |
| FindingSignificant at the nominal threshold only, since there was no correction for multiplicity. Not significant with multiple imputation (OR 1.635; 95% CI 0.877 to 3.049; p = 0.12) or in the per-protocol population (OR 1.671; 95% CI 0.883 to 3.162; p = 0.12). | |
| Response, week 8 | 67.8% vs 60%; adjusted OR 1.406 (95% CI 0.759 to 2.605); p = 0.28 |
| FindingNo difference shown. This does not prove the absence of an effect: the sample size had been calculated for a response difference of 46.6% vs 26.6%, and response on placebo reached 60%. | |
| Response, week 4 | OR 2.243; p = 0.009 |
| FindingSecondary outcome. | |
| Remission, week 4 | OR 2.701; p = 0.004 |
| FindingSecondary outcome. | |
| MADRS, week 4 | Difference of -4.37 points (95% CI -6.65 to -2.09); p = 0.0002 |
| FindingDifference in adjusted means, favouring perospirone. Secondary outcome. | |
| Placebo group | Response 60%; adverse events in 40.2% |
| FindingA placebo arm with this level of response leaves little room to show a difference. | |
Critical appraisal
| Item | Judgement |
|---|---|
| Primary outcome | Fragile |
| FindingOne primary outcome not significant, the other positive at the nominal threshold without correction and not confirmed by the sensitivity analyses. The authors describe the result as hypothesis-generating, while presenting perospirone as an option worth considering in their discussion. | |
| Early effect | Secondary signal |
| FindingThe separation at four weeks is clearer than at eight, but it rests on secondary outcomes. | |
| Blinding | Not tested |
| FindingAdverse events in 40.2% of patients on placebo; functional unblinding remains possible and was not measured. | |
| Missing data | Partly handled |
| FindingAbout 20% dropouts. Last observation carried forward is used for the main analysis of both primary outcomes, in a population limited to 177 of the 210 patients; multiple imputation is used only as a sensitivity analysis, and the mixed model applies to score changes. | |
| Transferability | Not listed in France |
| FindingWhether the drug can be prescribed depends on national marketing authorisations. In France, for example, no medicinal product containing perospirone is listed in the public medicines database as of 2 October 2026. The study says nothing directly about the antipsychotics authorised in a given country. | |
Level of evidence
The level of evidence is that of a randomised trial (Oxford CEBM level 1b). Confidence is reasonable that improvement comes faster with perospirone during the first month, on secondary outcomes that remain significant in the sensitivity analyses.
It is low for a benefit at eight weeks: one primary outcome is not significant, the other is significant only in the absence of correction and is no longer so in the sensitivity analyses, and the authors acknowledge the lack of correction. A confirmatory trial would be needed.
The colleague test
What an experienced colleague would say if you presented this study in two minutes, between two consultations.
“I don’t have the drug. What I take from it is that four weeks into a well-dosed SSRI that isn’t working, I don’t necessarily need to wait until week eight to augment. But one primary outcome that isn’t significant and a remission at p = 0.032 without correction do not amount to proof.”
Translation for practice: the question of when to augment deserves to be asked early, with the drugs available to you and within their authorised indications.
What you can do with this
- What you can do: formally reassess response four weeks into an antidepressant at an effective dose, rather than letting things drift to week eight without a decision.
- What you can understand: in this trial, augmentation with perospirone goes with faster improvement on secondary outcomes, but it does not demonstrate a benefit at eight weeks.
- What you can transfer, with caution: the principle of early augmentation, with the antipsychotics authorised where you practise and within their authorised indications, to be checked against your national register. Being indicated is a matter of authorisation; it does not settle whether a product is marketed or reimbursed. In France, for example, prolonged-release quetiapine is indicated for major depressive episodes in major depressive disorder, as add-on to another medicine prescribed for this condition; aripiprazole is not indicated for this use; brexpiprazole is not a component of any product in the database (public medicines database, consulted on 2 October 2026).
- What you can teach: how two uncorrected primary outcomes, a 60% placebo response and untested blinding change the meaning of an apparently positive trial.
- The course of action is set out in the NICE decision tree for depression in adults.
Frequently asked questions
Can perospirone be prescribed?
That depends on the marketing authorisations of the country where you practise. In France, for example, no medicinal product containing perospirone is listed in the public medicines database, consulted on 2 October 2026, so this trial cannot directly change a prescription there. Its interest lies elsewhere: the timing of augmentation and the critical reading of the trial.
Does the trial show that perospirone works as an add-on treatment?
Not in a confirmatory way. Remission at eight weeks is more frequent at the nominal threshold, but no longer so in the sensitivity analyses; the difference in response is not significant, and no correction was made for these two outcomes. The authors describe the result as hypothesis-generating.
Should an antidepressant be augmented as early as four weeks?
The trial suggests faster benefit when augmenting at that point, on secondary outcomes. It does not compare early with late augmentation, which would be the real question. Nearly a quarter of patients (24.86%) had received more than four weeks of antidepressant; in the post hoc analysis restricted to those treated for exactly four weeks, the difference at eight weeks is smaller (adjusted ORs 1.368 for response and 1.488 for remission, not significant), with no significant interaction between treatment and subgroup.
Why is a 60% placebo response a problem?
Because it leaves little margin to show a difference. A high placebo response reduces the power of the trial and makes a negative result hard to interpret.
Annotated bibliography
Source study. Liu J, Gao S, Liu B, Ju Y, Liao M, Zhang L, Li Z, Zhang Q, Zhao X, Tang H, Jiang R, Zhou J, Yan D, Wang Z, Zheng Y, Cai D, Wei N, Zhang Y, Yu H, Ding L, Wang B, Liu G, Li K, Hu S, Yang F, Kuang L, Yao Z, Zhang Y, Li L. Efficacy and safety of perospirone as adjunctive therapy in major depressive disorder patients with inadequate response to antidepressants: a randomized clinical trial. eClinicalMedicine. 2025;90:103626. DOI: 10.1016/j.eclinm.2025.103626. PMID: 41497519. Funding: STI2030-Major Projects (grant 2021ZD0202000), National Natural Science Foundation of China (grants 82201693 and 82471555), Hunan Provincial Natural Science Outstanding Youth Foundation (grant 2025JJ20093), Science and Technology Innovation Program of Hunan Province (grant 2024RC3057), Sanming Project of Medicine in Shenzhen (grant SZSM202311025) and Reboscience (Zhuhai) Pharmaceutical Research Co., Ltd.; trial sponsored by Livzon Pharmaceutical Group Inc., which supplied the investigational medicinal product and placebo; statistical analysis provided by Reboscience. Declared competing interests: none (“We declare no competing interests”); according to the authors, the funders and sponsor had no role in the design and conduct of the study, the collection, management and interpretation of the data, the preparation, review or approval of the manuscript, or the decision to submit it.
Context. Lam RW, Kennedy SH, Adams C, et al. Canadian Network for Mood and Anxiety Treatments (CANMAT) 2023 Update on Clinical Guidelines for Management of Major Depressive Disorder in Adults: réseau canadien pour les traitements de l’humeur et de l’anxiété (CANMAT) 2023 : mise à jour des lignes directrices cliniques pour la prise en charge du trouble dépressif majeur chez les adultes. Canadian Journal of Psychiatry. 2024;69(9):641-687. DOI: 10.1177/07067437241245384. PMID: 38711351. Canadian guideline cited by the authors (reference 5 of the source study) in support of reassessing the antidepressant after four weeks and of possibly adding a second-generation antipsychotic in case of inadequate response.
Editorial collections
Tags
Verified on 2 October 2026 against the full text of the publication and its supplementary material where available. This analysis underwent an independent double reading. The English version was checked for conformity on 2 October 2026, against the figures of the French version and against the source. How we verify what we publish
This analysis is intended for healthcare professionals. It does not constitute a prescribing recommendation and does not replace individual clinical judgment.
Analysis from Psychiatry Evidence Base, evidence-based psychiatry, explained with rigor.