Determinants of survival and retrospective comparisons of 183 clinical trial patients with myelofibrosis treated with momelotinib, ruxolitinib, fedratinib or BMS- 911543 JAK2 inhibitor

Between October 2007 and July 2013, 183 Mayo Clinic patients (median age 65 years; 58% males) with high/intermediate risk myelofibrosis (MF) were enrolled in consecutive phase 1/2 JAK2 inhibitor (JAKi) clinical trials with momelotinib (n = 79), ruxolitinib (n = 50), fedratinib (n = 23) and BMS-911543 (n = 31). Using conventional criteria, the respective response rates for spleen and “transfusion-dependent anemia” were 47%, 32%, 83%, 62% and 51%, 30%, 10%, 44%, respectively, favoring momelotinib for anemia response (p = 0.02) and fedratinib for spleen response (p < 0.01). All study patients were followed to death or 2022, during which time 177 (97%) drug discontinuations, 27 (15%) leukemic transformations, and 22 (12%) allogeneic stem cell transplants (ASCT) were recorded. 5/10-year survival rate for all 183 patients was 41%/16% and not significantly different across the four drug cohorts (p = 0.33). Multivariable analysis of pre-treatment variables identified age >65 years (HR 3.5), absence of type 1/like CALR mutation (HR 2.8), baseline transfusion need (HR 2.1), and presence of ASXL1/SRSF2 mutation (HR 1.6) as risk factors for overall survival; subsequent HR-based modeling segregated three risk categories with 5/10-year survival rates of 84%/60%, 44%/14%, and 21%/5% (p < 0.01). In addition, spleen (p < 0.01) and anemia (p = 0.01) responses were independently associated with improved short-term survival while long-term survival was secured only by ASCT (5/10-year survival rate 91%/45% vs 47%/19% in non-transplanted patients; p < 0.01). The current retrospective study suggests the value of specific pre-treatment variables in identifying long-lived MF patients receiving JAKi and also confirms recent observations on the favorable impact of treatment response on short-term and of ASCT on long-term survival.

The favorable impact of JAKi on splenomegaly and constitutional symptoms, and of momelotinib on anemia in patients with MF is well-established, but whether achievement of response following JAKi therapy positively influences long-term survival is unclear. Accordingly, in the current study, which includes MF patients treated on JAKi clinical trials at the Mayo Clinic, our primary objective was to i) retrospectively compare long-term treatment outcomes in momelotinib, ruxolitinib, fedratinib, and BMS-911543 treated patients with MF, ii) identify clinical and molecular predictors of response, overall and leukemia-free survival, and iii) examine the impact of treatment response on survival.

METHODS
The current study includes JAKi naïve patients with primary myelofibrosis (PMF), post-polycythemia vera and post-essential thrombocythemia MF enrolled in consecutive phase-1/2 momelotinib (NCT00935987), ruxolitinib (NCT00509899), fedratinib (NCT00631462, NCT01420770), and BMS-911543 (NCT01236352) clinical trials between October 2007 and July 2013. The latter clinical trial was terminated by the sponsor in November 2015. Study patients were retrospectively recruited after institutional review board approval with follow-up updated in July 2022. Patient eligibility, study design, drug doses and schedule have previously been published [2,5,9,10]. Diagnosis of PMF, post-ET and post-PV MF were according to conventional criteria [12]. Baseline transfusion status, spleen and anemia response were assessed according to the modified revised international working group for myelofibrosis research and treatment (IWG-MRT) criteria [13]. Screening for JAK2, MPL, CALR, ASXL1, SRSF2, IDH1/2, and U2AF1 mutations and dynamic international scoring system (DIPSS)-plus risk stratification and unfavorable karyotype categorization were as previously described [14,15]. All statistical analyses considered clinical and laboratory variables obtained at the time of study initiation. Conventional statistical methods were applied for response and survival analysis. Survival was calculated from the time of study drug initiation to last follow-up or death and survival curves prepared by the Kaplan-Meier method and compared by the log-rank test. Cox proportional hazard model was used for multivariable analysis. JMP Pro 16.0.0 software package, SAS Institute, Cary, NC was used for statistical analysis.  likely to present with leukocytosis (p = 0.02) and thrombocytosis (p = 0.02) while patients receiving BMS-911543 were more likely to be females (p = 0.04) and affected by constitutional symptoms (p < 0.01). However, ASXL1/SRSF2 mutation distribution and median time from diagnosis to initiation of therapy was similar for each treatment group (p > 0.1).

DISCUSSION
We pioneered early clinical development of momelotinib, ruxolitinib, fedratinib and BMS-911543 JAKi for the treatment of MF, therefore the current study is uniquely poised to provide comparative data on long-term outcomes for patients treated with each JAKi [2,5,6,9,10]. The vast majority of patients were followed until death, and follow-up for living patients exceeded 14 years. The findings from the current study confirm the palliative role of JAKi therapy through reduction of splenomegaly, achievement of transfusion-independence and relief of constitutional symptoms [16]. In that regard, a novel retrospective comparison of anemia and spleen response across all four JAKi is unveiled; fedratinib was associated with higher spleen response, while anemia response was superior with momelotinib. In addition, spleen response was found to be superior in the absence of ASXL1/SRSF2 mutations. Previous studies on the impact of mutations on spleen response in JAKi treated patients with MF have yielded conflicting results. In an analysis of MF patients treated with ruxolitinib in the COMFORT-2 study, spleen response was not influenced by mutational profile [17], while another study showed superior spleen response in the absence of ASXL1/EZH2/ IDH1/2 mutations in ruxolitinib treated patients [18]. On the other hand, we have previously reported superior spleen and anemia response in the absence of ASXL1 mutations in momelotinib treated patients with MF [19,20]. Additional noteworthy observations include high treatment discontinuation rates, with less than 5% of study patients remaining on long-term therapy (>10 years). Importantly, twothirds of patients discontinued therapy due to suboptimal response or disease progression, while death or leukemic transformation on-study was documented in a minority (<5%) of patients. In our series, the high discontinuation rates secondary to toxicity was likely a result of exposure to higher drug dosage in the phase-1 studies. The high treatment discontinuation rates attest to the transient benefit from JAKi therapy and although short-term survival was superior in spleen and anemia responders, long-term survival was not impacted by neither achievement of spleen nor anemia response with 10-year survival rates of <25%. Moreover, we have recently shown that the short-term survival benefit associated with anemia response in momelotinib treated MF patients might be limited to those with unfavorable genetic profile [21]. It is to be noted that patients remained on momelotinib longer and had higher anemia response rates compared to ruxolitinib, despite which survival was similar, which is likely due to momelotinib treated patients being older and belonging to higher DIPSS-plus risk.
The subject of whether MF patients derive long-term survival benefit from JAKi therapy remains contentious. A pooled analysis of the COMFORT-1 and 2 studies and a recently published prospective real-world series suggested prolonged survival in ruxolitinib-treated patients [22,23]. In contrast, patients treated on momelotinib and ruxolitinib clinical trials at the Mayo Clinic did not show a significant survival advantage when compared to risk-adjusted patient cohorts not receiving JAKi therapy [19,24]. The current study reinforces the prognostic impact of previously established clinical and molecular risk factors for survival in MF patients treated with JAKi, namely age >65 years, absence of type 1/like CALR mutation, presence of ASXL1/SRSF2 mutations, and transfusion dependence [15,25]. Based on the aforementioned variables, we propose a practical three-tiered survival prediction model for JAKi-treated MF patients, which enables identification of long-lived patients with 10 year survival of 60%. On the other hand, MF patients categorized as "high-risk" had considerably shortened long-term survival with 10-year survival of 5%, despite receiving JAKi therapy. Although achievement of anemia and spleen response improved short-term survival, a durable survival benefit was secured only by ASCT, which could have possibly resulted from ASCT candidacy itself by favoring younger patients with fewer medical comorbidities. Limitations of the current study include the retrospective design, which resulted in our inability to account for the impact of subsequent therapies on survival. Taken together, our findings underscore the transient benefit derived from JAKi therapy in MF in terms of palliation of splenomegaly and constitutional symptoms, however, impact on long-term survival was limited and primarily determined by pre-treatment clinical and molecular risk profile. Accordingly, early ASCT referral for patients with higher risk MF is advised.

DATA AVAILABILITY
Please email the corresponding author.