Luvadaxistat Fails To Improve Cognition in ERUDITE Phase 2 Trial

Key Takeaways

• Neither the 20 mg nor 50 mg dose of luvadaxistat improved the primary cognitive endpoint versus placebo at Day 98.
• Secondary cognitive, functional, and clinical measures also did not differ significantly from placebo.
• Adverse events affected about one-third of participants across groups, were mostly mild or moderate, and no major safety signal was reported during the trial period.

ERUDITE was a phase 2 randomized, double-blind, placebo-controlled, parallel-group study in adults aged 18–50 years with schizophrenia who were receiving stable antipsychotic therapy. Participants had stable symptoms and a reliable informant, then entered a 2-week placebo run-in before the 12-week double-blind period. Investigators randomized 203 participants in a 2:1:1 ratio to placebo, luvadaxistat 20 mg, or luvadaxistat 50 mg once daily and 178 participants, or 87.7%, completed treatment. Conducted across 55 sites in Europe and the USA between December 2021 and October 2024, the primary endpoint was change from baseline to Day 98 in BACS composite score.

Neither dose significantly improved the primary endpoint versus placebo on the Day 98 BACS assessment. Least-squares mean differences were −0.7 (95% CI −2.8 to 1.4; p=0.75) for 20 mg and −0.5 (95% CI −2.7 to 1.6; p=0.69) for 50 mg versus placebo. Because multiplicity adjustment was not applied across doses and endpoints, the p values were nominal. No significant differences were observed in cognitive functioning compared to placebo based on SCoRS differences, which were −0.2 for 20 mg and −0.6 for 50 mg. VRFCAT, CGI-S, PANSS total score, and BACS domains also showed no statistically significant separation from placebo. No broader efficacy signal emerged across cognition, function, or symptom measures.

Treatment-emergent adverse events occurred in about one-third of participants across groups, and most events were mild or moderate. No deaths occurred, and serious TEAEs were uncommon, affecting no more than 3% of participants in any group. Discontinuations due to TEAEs occurred in four placebo participants and one participant receiving luvadaxistat 50 mg, while somnolence and influenza were among the more frequent events. Increases in suicidal ideation appeared in three placebo participants and one 50 mg participant, with none in the 20 mg group. Small QTcF increases occurred in a few luvadaxistat recipients, but none exceeded 500 ms and no clinically significant ECG abnormalities were identified. Overall safety findings were broadly comparable across groups during the double-blind treatment phase.

The authors noted that ERUDITE did not reproduce the exploratory cognitive signal previously seen in the INTERACT program. They also highlighted interpretive limits, including nominal p values without multiplicity adjustment, possible population heterogeneity, and variability in cognitive performance before randomization. Lack of enrichment based on pre-baseline cognitive performance may also have influenced interpretation of the null results, and missing data were handled with placebo-based control multiple imputation. The authors also noted that luvadaxistat development had been discontinued within the broader program.