Highlights
– The CONNEX programme (three identical phase 3 trials) found no significant improvement in cognition with 10 mg iclepertin versus placebo at 26 weeks (MCCB overall composite T-score adjusted mean difference 0.127; 95% CI -0.396 to 0.650; p=0.63).
– Iclepertin was well tolerated with on-treatment adverse event rates similar to placebo (66% vs 69%); no safety signal emerged.
– The negative result underscores persistent challenges in CIAS drug development: patient heterogeneity, endpoint sensitivity, placebo/practice effects, and uncertain target engagement.
Background: the unmet need in cognitive impairment associated with schizophrenia (CIAS)
Cognitive impairment is a core and disabling feature of schizophrenia that affects attention, memory, working memory, processing speed, and executive function. These deficits are a major determinant of functional outcome — vocational performance, independent living, social functioning — and respond poorly to antipsychotic treatment. Despite decades of research, there are currently no regulatory-approved pharmacotherapies specifically for cognitive impairment associated with schizophrenia (CIAS). This therapeutic gap has driven interest in agents that modulate glutamatergic neurotransmission, including glycine transporter-1 (GlyT1) inhibitors designed to indirectly enhance N-methyl-D-aspartate (NMDA) receptor function by increasing synaptic glycine concentrations.
Study design: the CONNEX programme at a glance
The CONNEX programme comprised three parallel, phase 3, randomized, double-blind, placebo-controlled trials (CONNEX-1, CONNEX-2, CONNEX-3) conducted across 338 specialist psychiatric centres in 41 countries. Adults aged 18–50 years with schizophrenia were randomized 1:1 to oral 10 mg iclepertin once daily or matched placebo, each added to standard-of-care antipsychotic therapy, and treated for 26 weeks. The primary efficacy endpoint was change from baseline to week 26 on the Measurement and Treatment Research to Improve Cognition in Schizophrenia (MATRICS) Consensus Cognitive Battery (MCCB) overall composite T-score — a widely used, standardized cognitive outcome in schizophrenia trials.
Trial procedures included masking of patients, investigators, and personnel. Efficacy analyses included randomized patients who received study treatment; safety analyses included participants who took at least one dose. Individuals with lived experience contributed to trial design and outcome selection. Trials were registered on ClinicalTrials.gov (NCT04846868, NCT04846881, NCT04860830) and funded by Boehringer Ingelheim.
Key findings
Participants and baseline characteristics
Between September 8, 2021, and April 24, 2024, 1,835 patients received at least one dose of trial medication (918 iclepertin, 917 placebo); 1,602 completed the 26-week treatment period. Overall mean age was 34.7 years (SD 8.8); 66% were male. Most participants were White (49%) or Asian (30%); 67% were not of Hispanic or Latino ethnicity. Baseline cognitive impairment consistent with schizophrenia was present, as reflected by MCCB baseline scores (population means were not detailed in the summary but were consistent with CIAS cohorts typical for MCCB studies).
Primary outcome
At week 26 there was no significant difference between groups on the MCCB overall composite T-score. The adjusted mean difference (iclepertin vs placebo) was 0.127 (95% CI -0.396 to 0.650; p=0.63). Interpreting this effect in context: MCCB T-scores are standardized with mean 50 and SD 10, so the observed point estimate corresponds to an effect size of approximately 0.013 SD — essentially zero and far below thresholds generally considered clinically meaningful.
Secondary and exploratory outcomes
The Lancet Psychiatry summary reports the primary endpoint as negative; details of prespecified secondary cognitive domains, functional measures, and subgroup or biomarker analyses are not provided in the summary available here. In phase 3 programmes of this size, investigators typically prespecify multiple secondary and exploratory outcomes (individual MCCB domain scores, functional capacity scales, composite functional endpoints); such analyses may be published in full in the trial report and supplementary material. The available summary emphasizes the null primary result and tolerability profile.
Safety and tolerability
Iclepertin was generally well tolerated. On-treatment adverse events were reported in 66% (604/918) of iclepertin-treated patients and 69% (633/917) of placebo-treated patients. Serious adverse event and discontinuation rates were not highlighted as differing meaningfully between groups in the summary. One death occurred in the placebo group and was deemed unrelated to study treatment. No novel safety signal attributable to glycine transporter inhibition was described.
Interpretation and mechanistic considerations
The CONNEX programme’s negative primary outcome indicates that 10 mg iclepertin once daily did not produce clinically meaningful improvement in global cognitive performance versus placebo in adults with schizophrenia over 26 weeks, despite adequate power and rigorous methodology. This result mirrors prior disappointments in the field: several glutamatergic strategies, including other GlyT1 inhibitors, have failed to demonstrate robust, reproducible cognitive benefits in large trials.
Potential explanations span pharmacology, trial design, and disease biology. Mechanistically, GlyT1 inhibition aims to increase synaptic glycine to facilitate NMDA receptor signaling — a plausible target given NMDA hypofunction hypotheses in schizophrenia. However, bridging target engagement to clinical benefit can fail if dose or exposure does not achieve sufficient central GlyT1 inhibition, if downstream signaling adapts, or if NMDA receptor dysfunction in CIAS is not the rate-limiting factor for cognitive deficits in the treated population.
From a trial-design perspective, large cognition trials are challenged by high placebo and practice effects, heterogeneity in baseline impairment, and modest signal-to-noise ratios on neuropsychological batteries. The MCCB is standardized and validated, but even so, repeated testing and practice effects can attenuate detectable between-group differences, particularly in relatively younger or less impaired samples. Concomitant antipsychotic medications and concomitant psychosocial interventions (or lack thereof) may also modify the ability to detect pharmacologic benefit. Finally, patient selection criteria (age 18–50, chronicity, functional status) may matter; benefits could be more likely in specific subgroups (e.g., early-course illness, those with a biomarker-defined profile of NMDA hypofunction) that are diluted in broad enrollment.
Study strengths and limitations
Strengths include the large pooled sample size, multicentre/multinational execution, randomized double-blind placebo-controlled design, 26-week duration, standardized cognitive outcome (MCCB), and patient involvement in design. Safety data benefit from the large exposure pool and balanced groups.
Limitations include the lack of reported pharmacodynamic or biomarker data in the summary (plasma/CSF measures of glycine, PET or EEG markers of NMDA modulation), which would help determine whether central target engagement was achieved. The summary does not report detailed secondary outcome results, subgroup analyses, or functional endpoints; such data will be essential to fully understand whether any domain-specific or subgroup benefits exist. Finally, even well-conducted cognitive trials contend with practice effects and measurement noise that can obscure small treatment effects.
Implications for clinical practice and future research
For clinicians, CONNEX provides robust phase 3 evidence that 10 mg iclepertin as an adjunct to antipsychotics does not improve global cognition in adults with schizophrenia and therefore should not be adopted for this indication based on current data. The absence of a safety signal is reassuring but does not offset the lack of efficacy.
For researchers and sponsors, the programmes highlight priorities for future CIAS development: (1) incorporate biomarkers of target engagement and mechanistic readouts early and prospectively to confirm CNS pharmacology; (2) enrich trial populations by selecting patients more likely to benefit (biomarker-driven, early-stage illness, or specific cognitive profiles); (3) combine pharmacotherapy with validated cognitive remediation or rehabilitation interventions to attempt synergistic improvements in cognition and functioning; (4) optimize outcome measurement — consider alternate endpoints, hierarchical testing strategies, or adaptive designs to detect domain-specific effects; and (5) account explicitly for practice effects through alternate test forms, extended lead-in assessments, or statistical correction methods.
Conclusion
The CONNEX phase 3 programme found no clinically meaningful cognitive benefit of 10 mg iclepertin over placebo at 26 weeks in adults with schizophrenia, although the drug was well tolerated. These robust negative findings underscore how difficult it remains to translate mechanistically plausible targets into effective treatments for CIAS. Future success will likely require tighter integration of pharmacology, biomarkers, patient stratification, and combined behavioral-pharmacologic strategies.
Funding and trial registration
Funding: Boehringer Ingelheim. ClinicalTrials.gov identifiers: NCT04846868, NCT04846881, NCT04860830.
References
Keefe RSE, Harvey PD, Correll CU, Falkai P, Hashimoto N, Klein H, Krystal JH, Marder S, Medalia A, Sumiyoshi T, Wang G, Zhang H, Blahova Z, Bichard-Sall I, English BA, Fu E, Gruenenfelder F, Groeschl M, Kimura K, Tang W, von der Goltz C, Fowler C. Efficacy and safety of iclepertin for cognitive impairment associated with schizophrenia (CONNEX programme): results from three phase 3 randomised controlled trials. Lancet Psychiatry. 2025 Dec;12(12):906-920. doi: 10.1016/S2215-0366(25)00296-2. PMID: 41233083.
