Background
Multicentre, open-label, randomised, controlled phase 3 trial at five centres in China. 215 patients randomly assigned; 206 in the modified intention-to-treat population. Eligible patients were aged 12-70 years with haematological malignancies for which allogeneic HSCT was indicated, Karnofsky (or Lansky, if <16 years) performance status ≥70, undergoing their first myeloablative haploidentical HSCT. Median recipient age 40 years (IQR 24-48); 98 (48%) female; all participants Chinese. Ruxolitinib is a JAK1/2 inhibitor with established activity in steroid-refractory acute and chronic GVHD; the question here was whether replacing mycophenolate mofetil with low-dose ruxolitinib inside an ATG/calcineurin-inhibitor/methotrexate backbone reduces acute GVHD. NCT04838704.
Results
Interventions and follow up: Arm A: Antithymocyte globulin + calcineurin inhibitor + short-course methotrexate + low-dose oral ruxolitinib, started day 1 at 5 mg twice daily (≥50 kg) or 5 mg once daily (<50 kg), continued to day 60, then tapered to day 90 in the absence of grade II-IV acute GVHD (n=103)
Arm B: Antithymocyte globulin + calcineurin inhibitor + short-course methotrexate + mycophenolate mofetil (n=103)
Primary endpoint: Cumulative incidence of grade II-IV acute GVHD by day 100
mFollow up: 26.4 months (IQR 20.0-33.3) among surviving patients
Results: Grade II-IV acute GVHD by day 100: 7 patients, cumulative incidence 6.8% (95% CI 1.9-11.7) vs 38 patients, 36.9% (95% CI 27.5-46.3); subdistribution HR 0.15 (95% CI 0.07-0.34), P<.001
Grade III-IV acute GVHD: NR
Chronic GVHD: NR
Relapse: NR
GRFS: NR
OS: NR
Arm B: Antithymocyte globulin + calcineurin inhibitor + short-course methotrexate + mycophenolate mofetil (n=103)
Primary endpoint: Cumulative incidence of grade II-IV acute GVHD by day 100
mFollow up: 26.4 months (IQR 20.0-33.3) among surviving patients
Results: Grade II-IV acute GVHD by day 100: 7 patients, cumulative incidence 6.8% (95% CI 1.9-11.7) vs 38 patients, 36.9% (95% CI 27.5-46.3); subdistribution HR 0.15 (95% CI 0.07-0.34), P<.001
Grade III-IV acute GVHD: NR
Chronic GVHD: NR
Relapse: NR
GRFS: NR
OS: NR
Adverse events
Grade 3-4 haematologic: thrombocytopenia 17 (17%) vs 11 (11%); neutropenia 14 (14%) vs 9 (9%); anaemia 12 (12%) vs 9 (9%)
Grade 3-4 non-haematologic: cystitis 7 (7%) vs 10 (10%)
Serious adverse events: 31 (30%) vs 36 (35%)
Fatal adverse events: 0 in the ruxolitinib group; 3 deaths in the standard group (pulmonary infection, sepsis/bacteraemia/fungaemia, transplant-associated thrombotic microangiopathy) — none attributed to study drug
Grade 3-4 non-haematologic: cystitis 7 (7%) vs 10 (10%)
Serious adverse events: 31 (30%) vs 36 (35%)
Fatal adverse events: 0 in the ruxolitinib group; 3 deaths in the standard group (pulmonary infection, sepsis/bacteraemia/fungaemia, transplant-associated thrombotic microangiopathy) — none attributed to study drug
Conclusions
Substituting low-dose ruxolitinib for mycophenolate mofetil within an ATG, calcineurin inhibitor, and short-course methotrexate backbone reduced grade II-IV acute GVHD by day 100 from 36.9% to 6.8% after myeloablative haploidentical HSCT, an approximately six-fold relative reduction. Toxicity was manageable, with slightly more grade 3-4 cytopenias but fewer serious adverse events and no fatal events attributed to ruxolitinib. This is the first randomised phase 3 evidence supporting JAK1/2 inhibition as GVHD prophylaxis rather than treatment.
Key Limitations
Open-label design with an unblinded, clinically graded primary endpoint — acute GVHD grading is subject to assessor judgement, and the effect size (sHR 0.15) is unusually large for a prophylaxis intervention. Conducted at five centres in a single country with an entirely Chinese cohort, limiting generalisability. The comparator is an ATG/MMF backbone; it is not post-transplant cyclophosphamide, which is standard for haploidentical and mismatched transplant in much of North America and Europe, so this trial does not establish superiority over contemporary Western practice. The primary endpoint is a day-100 surrogate: chronic GVHD, relapse, GVHD-free relapse-free survival, and overall survival are not reported, and reducing early acute GVHD without a demonstrated relapse-free survival benefit is not by itself practice-changing. Modest sample size (206 mITT), and 9 randomised patients were excluded from the mITT population for major protocol deviations or not receiving prophylaxis. Investigator-initiated and academically funded (National Natural Science Foundation of China).
Clinical Context
Ruxolitinib is approved by FDA and EMA for steroid-refractory acute GVHD (REACH2) and for chronic GVHD after failure of one or two prior lines (REACH3); it is not approved for GVHD prophylaxis, and this trial does not change that status. In North America and Europe, post-transplant cyclophosphamide-based prophylaxis is the prevailing standard for haploidentical and mismatched unrelated donor transplant following BMT CTN 1703, while ATG-based backbones remain widely used in China and parts of Asia — which is the context these results speak to most directly. Confirmatory randomised comparison against a PTCy backbone, with GRFS, relapse, and survival endpoints, would be required before adoption outside an ATG-based platform.