Key Takeaways
- The 80 mg arm was stopped after interim futility analysis because it had a less than 20% probability of reaching an effect size of at least 0.35, supporting early discontinuation of underperforming CIAS doses in adaptive trial designs.
- At 24 weeks, the primary endpoint showed virtually no separation from placebo on the MCCB neurocognitive composite T-score, with mean changes from baseline of 1.08± 5.78 [SD] for placebo and 1.36±4.80 [SD] for 240 mg basmisanil (ES=0.05, P =.793).
- A signal on 1 memory measure should be interpreted cautiously: the delayed recall condition VPA total raw score II reached an ES of 0.45 (P<.1) at week 24, but there were no corresponding benefits on other verbal memory, working memory, global cognition, functional, or symptom measures.
- The prespecified subgroup strategy did not identify a clinically useful responder profile; numerical effects favored older patients (ES =0.39) over younger patients (ES=−0.32) and the preserved subtype (ES=0.27) over deteriorated (ES =0) or compromised (ES=−0.04) cognition.
- Tolerability was acceptable despite high overall attrition: 81 (37.9%) patients withdrew prematurely, but safety-related withdrawals were only 3.7% to 5.6%, and serious adverse events were similar in the 240 mg treatment (4 events) and placebo (5 events) groups.
- Psychosis worsening was not increased with basmisanil in this sample, occurring in 3.7% and 3.8% of the 80 mg and 240 mg groups versus 6.2% with placebo, with no observed proconvulsive liabilities, insomnia, anxiety, depression, sedation, or suicidality.
-
Do not expect basmisanil 240 mg twice a day to improve CIAS in routine practice; after 24 weeks, the MCCB neurocognitive composite T-score changed by 1.08± 5.78 [SD] with placebo and 1.36±4.80 [SD] with basmisanil (ES=0.05, P =.793).
-
An isolated cognitive signal is not enough to justify a procognitive claim; the delayed recall condition VPA total raw score II reached an ES of 0.45 (P<.1) at week 24, but no benefit emerged on verbal learning, working memory, other MCCB domains, functional measures, or symptom ratings.
-
Do not target younger patients or the deteriorated cognitive subtype for basmisanil, because the prespecified subgroup strategy failed: older patients showed ES =0.39, younger patients ES=−0.32, and subtype effects were ES=0.27 for preserved, ES =0 for deteriorated, and ES=−0.04 for compromised.
-
Repeated pre-randomization cognitive testing can suppress early placebo inflation in CIAS trials; with 3 MCCB administrations before randomization, no improvement in the placebo group was noted at week 12, and the second assessment showed a change of 1.08, below the 1.9 points cited from prior controlled trials.
-
High attrition does not necessarily mean poor tolerability in schizophrenia cognition trials; 81 (37.9%) patients withdrew prematurely, but safety-related withdrawals were only 3.7% to 5.6%, and worsening of psychosis was reported in 6.2% with placebo versus 3.8% with 240 mg basmisanil.
