Background: MMB, an oral JAKi, has been shown in early trials to reduce spleen volume, improve disease associated symptoms (Sx) and improve RBC transfusion (Tx) requirements in patients (pts) with MF. This study was designed to test non-inferiority of MMB vs RUX in splenic volume reduction and Sx amelioration, and superiority in Tx requirement, in JAKi naïve MF pts. Methods: Eligibility: MF, IPSS high risk, Int-2, or symptomatic Int-1; palpable spleen ≥5cm; platelets ≥ 50 K/μl, and no Gr ≥2 peripheral neuropathy (PN). Stratification by Tx dependency and platelets (<100, 100-200 and >200 K/μl). Pts were randomized 1:1 to 24 wks of MMB 200 mg qd + RUX placebo or RUX 20 mg bid (or modified per label) + MMB placebo, after which all pts could receive open label MMB. Assessments: spleen volume by MRI, and pt reported Sx using a daily eDiary of modified MPN-SAF Total Sx Score (TSS). Primary endpoint was splenic response rate (SRR; ≥35% reduction in volume from baseline) at 24 wks. Secondary endpoints, evaluated sequentially at 24 wks, were rates of TSS response (≥50% reduction from baseline), RBC Tx independence (TI), RBC Tx dependence (TD) and of RBC Tx.Results: 175 of 215 (81%) and 201 of 217 (93%) pts randomized to MMB and RUX, respectively, completed the 24 wk DB phase. Efficacy results are shown in Table. Most common Gr ≥3 AEs in the DB phase with MMB were thrombocytopenia (7%) and anemia (6%), and with RUX were anemia (23%), thrombocytopenia (5%) and neutropenia (5%). Gr ≥3 infections occurred in 7% of MMB and 3% of RUX pts. Treatment emergent PN occurred in 22 (10%) of MMB (all Gr ≤2) and 10 (5%) of RUX (9 Gr ≤2, 1 Gr 3) pts in DB phase, none discontinuing study drug for PN. Overall, AEs led to study drug D/C in 13% of MMB and 6% of RUX pts in DB phase. Conclusions: In pts with JAKi naive MF, 24 weeks of MMB is non-inferior to RUX for spleen response but not for symptom response. MMB treatment is associated with a reduced transfusion requirement. NCT01969838.

^aTest for non-inferiority; ^bTest for superiority, all values nominally significant.