Background
Phase III, double-blind, placebo-controlled RCT. 440 patients with previously untreated non-GCB DLBCL (centrally confirmed by Hans algorithm). Tucidinostat is an oral subtype-selective HDAC inhibitor (class I/IIb). Preclinical rationale: HDAC inhibition enhances BCL6 degradation and re-sensitizes ABC-DLBCL to immunochemotherapy.
Interventions and follow up
Arm A: Tucidinostat 20mg PO twice weekly on days 1–14 + R-CHOP q21d × 6 cycles, followed by tucidinostat maintenance 20mg twice weekly × 12 months
Arm B: Placebo + R-CHOP q21d × 6 cycles, followed by placebo maintenance × 12 months
Primary endpoint: Progression-free survival (investigator-assessed)
Median follow-up: 31.1 months
Arm B: Placebo + R-CHOP q21d × 6 cycles, followed by placebo maintenance × 12 months
Primary endpoint: Progression-free survival (investigator-assessed)
Median follow-up: 31.1 months
Results
PFS (ITT): Not reached vs 29.4 months, HR 0.63, 95% CI 0.46–0.86, P=.003
2-year PFS rate: 65.9% vs 55.0%
OS: Not reached vs not reached, HR 0.72 — immature, NR
ORR: 91.8% vs 85.5%
CR rate: 79.1% vs 68.2%
2-year PFS rate: 65.9% vs 55.0%
OS: Not reached vs not reached, HR 0.72 — immature, NR
ORR: 91.8% vs 85.5%
CR rate: 79.1% vs 68.2%
Adverse events
Overall: Grade ≥3 AEs 85.5% vs 72.3%; G-CSF use significantly higher in tucidinostat arm
Hematologic: Grade ≥3 neutropenia 70.0% vs 52.3%; thrombocytopenia 20.5% vs 5.9%; anemia 14.5% vs 5.9%; febrile neutropenia 20.9% vs 10.9%
Hematologic: Grade ≥3 neutropenia 70.0% vs 52.3%; thrombocytopenia 20.5% vs 5.9%; anemia 14.5% vs 5.9%; febrile neutropenia 20.9% vs 10.9%
Conclusions
Tucidinostat + R-CHOP significantly improved PFS in non-GCB DLBCL, representing the first positive Phase III trial of an HDAC inhibitor in this biomarker-selected population. This is a landmark result for a strategy that had failed with vorinostat and panobinostat in unselected DLBCL.
Key Limitations
OS data immature at primary analysis — PFS benefit may not translate to OS given post-progression salvage options. Trial conducted entirely in China, raising questions about external validity. Hans algorithm non-GCB classification is less precise than COO by GEP — some true GCB cases may be misclassified. Maintenance component makes it unclear whether induction or maintenance drives benefit. Substantial hematologic toxicity.
Clinical Context
Published in JAMA 2026 — very recent data. Tucidinostat is approved in China but not yet by FDA or EMA. If validated in non-Chinese populations, this could represent the first successful R-CHOP intensification strategy in biologically selected DLBCL, following consistent failures of ibrutinib (PHOENIX) and lenalidomide (ROBUST, SEXIE) in non-GCB disease. FDA approval pathway uncertain.