IWG-MRT 2013 Criteria-Based Assessment of Response Among 83 Patients with Myelofibrosis Treated with JAK Inhibitors: Experience of Two Centers

Introduction. Ruxolitinib is a JAK inhibitor (JAKi) approved for myelofibrosis (MF) patients (pts), while other JAKi are being tested in clinical trials. The IWG-MRT response criteria (RC) used to evaluate drug efficacy were revised on 2013 and adapted to take into account efficacy in specific aspects of the disease, duration of response and toxicities (Blood 2013;122:1395).

Methods. We retrospectively evaluated the response rate in MF pts treated for at least 3 months (mo) with JAKi in phase II/III and expanded access trials at University of Florence and Pavia, according to IWG-MRT 2013 RC. We also collected molecular data, last follow up and reasons of discontinuation. Changes in spleen size (SS) were evaluated only by palpatory speen length measurement from the left costal margin without confirmation by MRI or CT scan and symptomatic improvement evaluation was present/absent (no grading).

Results. We collected 83 pts, 66 treated with ruxolitinib (79.5%), 12 fedratinib (14.5%), 3 pacritinib (3.6%), 2 gandotinib/LY2784544 (2.4%). At enrolment, the 4 groups did not differ for gender, age, MF diagnosis (primary or secondary to polycythemia vera or essential thrombocythemia), DIPSS, DIPSS-plus, JAK2V617F, MPLW515 and CALR mutational status, and JAK2V617F allele burden. 64 cases (77%) were evaluated for high molecular risk status (HMR; at least one mutation in EZH2, ASXL1, IDH1/2 and SRSF2 genes): 22/64 (34.4%) were HMR with no differences among groups. Of the 3 pacritinib pts, one carried SRSF2 and one both EZH2 and ASXL1 mutation. At baseline, median hemoglobin (Hb, g/dL) levels were lower for fedratinib and pacritinib group: 11.5 (range 7-15.5) for ruxolitinib, 10 (7.4-13.3) for fedratinib, 8.1 (8-8.3) for pacritinib and 11.9 ( 11.9-12) for gandotinib (p=0.012). All pacritinib treated pts were red blood cell transfusion dependent, compared to 10.6% in ruxolitinib (7/66), 8.3% in fedratinib (1/12) and none in gandotinib (p<0.001). Median platelet count  (PLT, x10^9/L) was: 245 (range 52-603) for ruxolitinib, 153 (70-610) for fedratinib, 28 (27-119) for pacritinib, 170 (154-187) for gandotinib (p=0.065). Median SS was 14 cm (range 0-35) for ruxolitinib, 23 (10-36) for fedratinib, 20 (17-26) for pacritinib, 22 (21-23) for gandotinib (p=0.007).  9/66 pts treated with ruxolitinib (13.6%) and 5/12 with fedratinib (41.7%) had received previous therapy with another JAKi or combined therapy with other molecules. Median duration of treatment was 22 mo (range 1-64) with ruxolitinib, 7 mo (2-18) with fedratinib, 6 mo (4-7) with pacritinib and 4 mo (3-5) with gandotinib. No pts achieved IWG-MRT complete or partial response.  Clinical improvement (CI) was achieved by 23/64 evaluable pts with ruxolitinib (35.9%), 3/12 with fedratinib (25%), none with pacritinib and gandotinib.  Anemia response (AR) was obtained in 5/26 pts in ruxolitinib (19.2%), 1/6 in fedratinib (16.7%) and none in pacritinib group. Median duration of AR during ruxolitinib was 10 mo, range 5-60. Spleen response (SR) was obtained by 26/64 (40.6%) pts in ruxolitinib, 5/12 (41.7%) in fedratinib, none with pacritinib and gandotinib. Median duration of SR was 13.5 mo (range 4-63) with ruxolitinib and 8 mo (range 4-16) with fedratinib; SR rate with fedratinib might be underestimated due to premature interruption of the trial. Symptoms resolution was achieved by 25/27 pts with ruxolitinib (92.6%), 7/10 with fedratinib (70%) and 0/3 with pacritinib (p=0.0001). 46 patients discontinued JAKi due to adverse events or toxicities (n=17, 37.0%), study closure and inefficacy (n=9 each, 19.6%), disease progression (n=7, 15.2%) and consent withdrawn (n=4, 8.6%). SR was correlated with smaller spleen size at baseline: 63.3% versus (vs) 38.7% for less or more than 50^th percentile (16 cm, p=0.03) and 36.7% vs 9.7% for less or more than 75^thpercentile (22 cm, p=0.007) and with treatment duration: pts who obtained SR had a longer treatment duration (22 mo, range 4-64, vs 14 mo, range 3-62, p=0.028). Pts receiving a previous JAKi vs not pretreated pts had less SR (14 vs 43%, p=0.04) but also had larger SS at enrolment (p=0.021). 

Conclusions: In this series, CI frequency obtained with ruxolitinib was 35.9% and with fedratinib of 25%, with limitations due to qualitative only symptoms evaluation. SR and AR was obtained with similar frequencies with ruxolitinib and fedratinib. SR is influenced by smaller SS at treatment beginning and by longer treatment duration.