Comparison of clinical outcomes between transplant and nontransplant therapies in myelofibrosis following failure of first-line JAK-inhibitor

Following Food and Drug Administration (FDA) approval of ruxolitinib in 2011, medical management of symptomatic MF with JAK-inhibitor (JAKi) therapy has become the standard of care. JAKi therapy does not have clear impact on modifying disease course or improving survival and clinical benefit is variable with most patients experiencing JAKi therapy failure by 3 years with subsequent poor survival of 11–16 months.1 Reasons for JAKi failure are varied, with proposed Canadian MPN Group criteria serving as a guide to classify the pattern of failure.2 The FDA approval of additional JAKi, fedratinib and pacritinib, along with the development of novel therapies has broadened the available options; but it is not clear if second-line therapies have improved the poor prognosis. At our center, our approach has been to consider the option of allogeneic hematopoetic cell transplantation (HCT) in all eligible patients with failure of first-line JAKi therapy in a shared-decision model.3 There is limited information on efficacy of HCT in comparison to nontransplant therapies in patients following first-line JAKi therapy failure. A prior study based on a small number of patients following ruxolitinib discontinuation suggested improved survival in patients undergoing HCT, though did not adjust for confounding factors that determine fitness for transplant.4 To address this knowledge gap, we sought to compare outcomes in transplant-eligible patients with MF between those who underwent HCT and patients managed with best available nontransplant therapies (BAT) following first-line JAKi therapy failure. We conducted a retrospective review of patients with MF from a prospective MPN registry (NCT02760238) treated at the Princess Margaret Cancer Centre in Toronto, Canada between January 1, 2011 and February 28, 2022. Patients were included if they had a diagnosis of chronic phase Primary (overt or prefibrotic), (MF) or MF following previous diagnosis of essential thrombocythemia or polycythemia vera and experienced first-line JAKi failure as per definitions from the Canadian MPN Group (Table S1).2 Patients were excluded from study if they had progressed to an accelerated (10%–19% peripheral/bone marrow blasts) or blast phase (≥20% blasts) or were not candidates for HCT (Table S2). Patients for whom a hematopoietic cell donor was not found or declined HCT based on preference but were otherwise medically fit for transplant were included in the best-available therapy (BAT) group. Baseline clinical and laboratory data at the time of JAKi failure were collected. We used the age-adjusted Dynamic International Prognostic Scoring System (aaDIPSS) to better estimate disease-related risk given the younger age of the study cohort.5 Patients were analyzed with an intention-to-treat basis according to the main second-line therapy received: either HCT or BAT. The HCT group included patients who received bridging therapy (second-line JAKi or splenectomy) with the intent of optimization prior to HCT. The BAT group included patients who received second-line therapy with alternative JAKi or clinical trial novel agent (alone or in combination with JAKi); or best supportive care with hydroxyurea, transfusion support, danazol, growth factors (recombinant erythropoietin), splenectomy, or continuation of first line JAKi. The primary outcomes of interest were overall survival (OS) from time of JAKi failure to last follow-up or death. A key secondary outcome is transformation-free survival (TFS) defined as the time from JAKi therapy failure to last follow-up, death, or transformation AP/BP. Statistical methods are described in the Appendix A. Of the 641 patients with MF in our database, 323 (50.4%) were treated with first-line JAKi and 122 under the age of 70 years met the definition of JAKi therapy failure. Of these patients, 88 patients met the study eligibility criteria (Figure S1), and outcomes were compared between the HCT (n = 41), and BAT (n = 47) groups. Baseline variables are described in Table S3. The median [range] follow-up from time of JAKi failure in survivors was 39 [6–126] months. Of the 47 patients who received BAT, 38 (81%) patents had donor searches initiated and 35 had a donor identified. Second line treatments in the BAT arm were: fedratinib (n = 7), momelotinib (n = 6), ruxolitinib (n = 3), pacritinib (n = 3), investigational agent added to ruxolitinib (n = 3), continuation of first-line JAKi (n = 18), splenectomy (n = 6), and splenic radiation (n = 1). Two patients in the BAT group underwent salvage HCT following failure of second-line treatment. In the HCT group, 8 (20%) patients received second-line bridging treatment prior to HCT: fedratinib (n = 4), ruxolitinib after first-line momelotinib (n = 2), and splenectomy (n = 2). Compared to the BAT cohort, the HCT cohort was younger, had smaller median palpable spleen size, less frequent constitutional symptoms, and higher proportion of low/int-1 aaDIPSS. There were no observed significant differences in blood counts, peripheral blasts, transfusion needs, ECOG status, or pattern of first-line JAKi therapy failure. At the last follow up, 53/88 (60%) of the patients had died including 56% (23/41) in the HCT group and 64% (30/47) patients in the BAT group. The causes of death in the HCT group were transplant-related complications (15/23), relapsed/progressive MF (5/23), and other causes (3/23). In the BAT group, the most common cause of death was progression of MF, either in chronic-phase (13/30) or after transformation to AP/BP disease (12/30); with remaining deaths due to comorbid conditions. Median [95% confidence interval] OS was 46 [23–98] months in the HCT group and 24 [16–40] months with BAT (Figure 1). In total 2 (5%) and 15 (32%) patients in the HCT and BAT groups respectively progressed to an AP/BP; with median TFS of 51 [23–89] and 22 [14–34] months. Survival rates in two arms were similar in the first 12 months after JAKi failure but began to diverge in favor of HCT after the first year. Univariate analysis is summarized in Table S4. In multivariable models (Table S5) there was a benefit for patients managed with second-line HCT observed beyond 12 months following JAKi-therapy failure after adjustment for hemoglobin < 100 g/L and WBC >25 × 109/L (Model 1: OS HR0.37 [95%CI 0.17–0.79]; TFS HR0.37 [95%CI 0.17–0.79]), or aaDIPSS category (Model 2: OS HR0.44 [95%CI 0.20–0.95]; TFS HR0.44 [95%CI 0.21–0.98]). In this study, we report improved clinical outcomes with HCT for a meticulously constructed cohort of MF patients with failure of first-line JAKi therapy who are otherwise fit for HCT. Although outcomes within the first year following JAKi therapy failure are similar between treatment approaches, the long-term survival favored the HCT group after adjusting for predictive covariates. The study highlights that many patients decide against HCT despite high-risk disease and potential donors. Consistent with previous work by our group evaluating transplant decisions in MF, patients who proceed with HCT were more likely to have a matched sibling donor, while patients managed with BAT more often had haploidentical or other alternative donors identified.6 Compared to a previously published study, our study has several strengths.4 We did not compare outcomes among all patients that discontinued ruxolitinib as many would not be considered eligible for HCT due to age or comorbidity making direct comparisons to an HCT-eligible population prone to potential confounding. Our patients were identified from a prospective MPN registry and patients with failure of JAKi therapy are managed with a uniform approach that prompts re-evaluation for HCT. Our center has extensive experience with HCT in MF, as well as access to novel agents through several clinical trials were available to our patients. Despite screening a large registry of MF patients to construct our study cohort, only a relatively small proportion could be reasonably considered for the option of HCT at time of first-line JAKi therapy failure due to age or comorbid medical condition. Use of ruxolitinib discontinuation in prior studies as a proxy for time of failure can be misleading, as prior to approval of fedratinib in 2019, patients experiencing disease progression on ruxolitinib had limited options and often continued the drug despite clinical failure. Standardized criteria allow for recognition of treatment-failure and prompt evaluation for HCT in the second-line. Our study is limited by retrospective, single-center design and relatively limited sample size. Therefore, these observations should be validated in a large, multicenter cohort. Ideally, a comparison of HCT and nontransplant therapies would be conducted with a prospective study to compare outcomes following JAKi-failure; but practical and logistic challenges make the feasibility of such a study difficult. Within the BAT group, the most common causes of death were related to progression of MF, with up to 40% progressing to AP/BP. The most common therapies used as BAT were second-line JAKi or continuation of first-line JAKi with addition of supportive measures including transfusions, splenectomy, or growth factor support. JAKi therapy alone has not demonstrated significant disease-modifying effects to prevent progression of the disease. Development of novel agents with disease-modifying potential may lead to improved outcomes following first-line JAKi failure with future studies needed. In conclusion, our study demonstrates the superior long-term outcomes with HCT compared to BAT in patients with MF after failure of first-line JAKi. These data suggest that HCT should be considered in all transplant-eligible patients with MF after failure of first-line JAKi therapy. The therapeutic landscape of MF is rapidly evolving and strategies to reduce transplant-related complications, especially for nontraditional/mismatched donor sources, along with development of novel agents with disease modifying activity are needed to improve outcomes of patients with first-line JAKi therapy failure. This work was supported by grants from the Princess Margaret Cancer Centre Foundation and the Elizabeth and Tony Comper Foundation to the MPN Program at the Princess Margaret Cancer Centre (VG). JTE, EGA, JAK, BG, VC, TN, KG, MBD, AB, HS, and AV report no conflicts or competing interests to declare. DM has received search support from Novartis, Celgene/BMS, PharmaEssentia, Takeda; honoraria from Novartis, Celgene/BMS; and has served in consultation or on the advisory board for Novartis and Pfizer. VG has received research funding through his institution and honoraria from Novartis and Incyte and has served on the advisory board of Novartis, Incyte, BMS-Celgene, Abb Vie, Sierra Oncology, Pfizer, Takeda, and Constellation Biopharma. The data that support the findings of this study are available from the corresponding author upon reasonable request. Data S1. Supporting Information Please note: The publisher is not responsible for the content or functionality of any supporting information supplied by the authors. Any queries (other than missing content) should be directed to the corresponding author for the article.