Impact of Symptom Benefit and Transfusion Response on Survival in Myelofibrosis Patients Treated with Pacritinib: PERSIST-2 Landmark Survival Analysis

Background: Pacritinib is a JAK1-sparing inhibitor of JAK2/IRAK1/ACVR1 that demonstrated symptom benefit measured by TSS (v2.0, excluding tiredness) vs best available therapy (BAT) in PERSIST-2, which enrolled cytopenic myelofibrosis patients with platelets ≤100×10 9/L. Recent datashowed that in patients with cytopenic MF, a ≥10% reduction in spleen volume at 12 weeks is associated with improved survival in patients treated with pacritinib ( Ajufo H, et al., J Clin Oncol;2023:14(16):Suppl). Surprisingly, this association was not confirmed in patients treated with BAT (82% of whom were on ruxolitinib). The association between other important measures of clinical benefit such as symptom burden reduction and anemia improvement on survival in patients treated with pacritinib has not previously been described. Methods: Patients randomized to pacritinib 200 mg BID and to BAT on PERSIST-2 were included if they were alive and on study at the start of the week 12 efficacy assessment period. Survival was assessed by Total Symptom Score (TSS; v2.0 excluding tiredness) reduction at various thresholds (≥50%, ≥20%, ≥10%, >0%) and transfusion independence (TI) response. TI was assessed among patients who required red blood cell (RBC) transfusion at baseline (within 90 days) and was defined, for the purposes of a landmark survival analysis, as the absence of RBC transfusions over any 12-week period from first dose through week 14 (allowing for a ±2-week window around week 12). Overall survival (OS) was estimated via the K-M method following the landmark approach; survival in responders vs non-responders was compared using the log-rank test. Results: Of the 59 patients treated with pacritinib who had a reduction of TSS≥10%, the median age was 67 years, 25% had a baseline high-risk DIPSS, median platelet count was 61 x10 9/L, median hemoglobin was 9.7 g/dL, and baseline TSS was 19.1. Reduction of TSS≥10% was associated with improved OS, with 1 death among 59 responders vs 4 deaths among 30 non-responders (hazard ratio [HR]=0.12, 95% confidence interval [CI]: 0.01, 0.79] P=0.021, Fig 1a). Similar to spleen volume reduction, the association between TSS response and OS weakened at more stringent TSS response thresholds (Fig 1c, e), with the BAT arm not showing an association between TSS response and OS at any threshold. (Fig 1b, d, f). In order to assess the relationship of specific TSS domains and survival , a landmark OS analysis was performed based on ≥20% reduction in phys